CN102932419B - A kind of data-storage system for the safety production cloud service platform towards industrial and mining enterprises - Google Patents

Abstract

A data storage system for a safe production cloud service platform for industrial and mining enterprises. A data storage layer is provided in the data acquisition equipment. Aiming at massive data from various data sources in industrial and mining enterprises, it is processed and generated with The theme data of the unified interface corresponding to different businesses, and store these theme data in the distributed file system. When there is a task request, according to different task requests, the data stored in the distributed file system is multi-node, Multi-task parallel computing and analysis, and display the analysis results according to different applications.

Description

A data storage system for a safe production cloud service platform for industrial and mining enterprises

technical field

The invention belongs to the field of automatic information collection and control of industrial and mining systems, and relates to a power system real-time data collection, processing and control system, in particular to an industrial and mining real-time data integrated processing system and design method based on cloud computing.

Background technique

Cloud technology has been widely used in the Internet, mainly including three different types, software as a service SaaS, platform as a service PaaS, infrastructure as a service IaaS. Among them, PaaS provides complete or partial application development that users can access, and SaaS provides complete and directly usable applications, such as managing enterprise resources through the Internet. The cloud technology in IaaS is mainly based on massive data management technology, massive data distributed storage technology, virtualization technology, and cloud computing platform management technology. Among them, massive data management technology and distributed storage technology are important components of data processing. Cloud computing needs to process and analyze distributed and massive data. Therefore, data management technology must be able to efficiently manage a large amount of data. Redundant storage ensures data reliability. Widely used data storage systems in cloud computing systems are Google's GFS and HDFS, an open source implementation of GFS developed by the Hadoop team.

In the actual production and operation of safe production in industrial and mining enterprises, there must be massive data (such as various safety production standards, monitoring data, education and training knowledge, etc.) Parallel processing on multiple machines to improve computing speed. The cloud platform manages and stores the data information accumulated in the actual production of safety production enterprises in a unified manner, and provides functions such as data-based prediction and abnormal detection for the production process, so as to realize the safe production of enterprises.

The data collection function of traditional industrial and mining enterprises is closely combined with the specific logical judgment function, which is usually completed by a relatively single hardware device. Therefore, different devices or systems need to be configured to complete different functions. These devices or systems with different functions have their own Independent data acquisition unit. Various data collections have problems such as cross-repeated collection, low utilization rate, inconsistent data and information content, and inconsistent time, forming an "information island" characterized by multiple vertical levels and multiple horizontal systems, which restricts the further development of information. Convergence and application, repeated collection and repeated transmission of data and information will inevitably lead to waste of various resources. Instruments in most industrial and mining enterprises are divided into three categories: protection, measurement and control, and metering. They all have their own data acquisition and processing units, which can be connected with the monitoring and display system for basic parameter measurement. But there are also the following technical problems:

(1) The accuracy of the instruments is different, resulting in inconsistent data and uneconomical cost.

(2) The different acquisition frequencies and algorithms of the instruments cause the same value to be different in different instruments, and the data is redundant and confusing.

(3) It is difficult to communicate between the system and equipment, and between equipment and equipment, and there is a lack of unified standards.

(4) The process data is missing. During the sampling process, a large amount of process data is generated. These data are valuable for later accident analysis or business expansion, but what we get from the instrument is all calculated after N times "Secondary Data", "Simplified Values". The problems brought about by this independent configuration, independent calculation, and independent function device are poor information sharing, low utilization rate and waste of hardware and software resources.

In order to solve the above-mentioned problems, especially the key problems such as separate data collection, data processing and calculation, and data sharing, it is necessary to establish a new framework and mechanism to realize one-time data collection of the same type of equipment, which can be used in the entire set Under the unified software application platform, cloud computing technology is used to realize various application functions such as measurement and control, protection, phasor measurement, and metering.

Contents of the invention

The purpose of the present invention is to provide a safety production cloud service platform system for industrial and mining enterprises to solve the problems of poor information sharing, low utilization rate, and waste of software and hardware resources caused by independent equipment configuration, independent calculation, and independent function devices. Realize one-time collection of the same type of equipment data, and multiple applications such as measurement and control, protection, phasor measurement, and metering.

The safety production cloud service platform system for industrial and mining enterprises includes:

1) Establish a safety production cloud service platform for industrial and mining enterprises. This platform integrates functions such as massive data processing supporting safety production services, safety production management business collaboration, and system management, and provides important technologies for safety production and government supervision of industrial and mining enterprises support.

2) Based on the multi-dimensional association rule analysis technology of safety accident events, it can analyze and dig out the frequent factors and potential laws that may lead to accidents and near-misses from massive production data, and establish a dynamic monitoring and early warning analysis of the effectiveness of safety production standards and the absence of standards This method provides a scientific basis for the government and industry regulators to formulate risk-based and reasonable operating standards.

3) The massive data processing method in the safety production of industrial and mining enterprises is proposed, including the support vector machine-based safety production anomaly detection technology and the knowledge discovery technology based on generalized rule reasoning, which can analyze and knowledge mine the massive production data, thereby Help enterprises improve accident prevention, early warning and emergency response capabilities, and improve enterprise safety production level.

4) Develop a safety production management tool set to facilitate the safety management, supervision, configuration, etc. of safety production in the industrial and mining industries, and realize rapid investigation and rapid response of enterprises.

Among them, the safety production cloud service platform developed for industrial and mining enterprises integrates the functions of massive data processing supporting safety production service cloud, safety production management business collaboration, and system management to realize real-time monitoring and early warning of enterprise production safety.

Specifically, a data storage system for an industrial and mining enterprise-oriented safety production cloud service platform, characterized in that the industrial and mining enterprise-oriented safety production cloud service platform system is characterized in that it includes a safety cloud service platform portal subsystem , system management and related tool set development subsystem, application service layer subsystem, virtual resource layer subsystem, platform service support layer subsystem, access and adaptation layer subsystem, security service resource subsystem, infrastructure service support layer Subsystem, the application service layer subsystem also includes a massive data processing system, which is composed of five major parts: data acquisition equipment, data integration equipment, data management equipment, standardized service interface equipment, and high-speed and reliable optical fiber communication network, forming a cloud computing-based The real-time data integrated processing system for industrial and mining enterprises, which is equipped with a data storage layer in the data acquisition equipment, processes the massive data from various data sources in industrial and mining enterprises, and generates unified interfaces corresponding to different businesses after processing them Subject data, and store these subject data in the distributed file system, when there is a task request, according to different task requests, perform multi-node, multi-task parallel computing and analysis on the data stored in the distributed file system , and display the analysis results according to different applications.

The data storage system for the safety production cloud service platform for industrial and mining enterprises is characterized in that: the data storage system includes at least one cloud storage management node, at least one cloud storage space and at least one virtual device, and the cloud storage management Nodes, cloud storage space and virtual devices constitute a private cloud.

The data storage system for the safety production cloud service platform for industrial and mining enterprises is characterized in that: the data storage system includes two cloud storage management nodes, three cloud storage spaces and multiple virtual devices, three cloud Storage Spaces and multiple virtual appliances form a private cloud.

The data storage system for the safety production cloud service platform for industrial and mining enterprises is characterized in that: the data storage system includes two cloud storage management nodes a, b, cloud storage management node a and cloud storage management node b Interconnected, the two cloud storage management nodes take over each other, the load can be balanced between the two cloud storage management nodes and one of the cloud storage management nodes takes over each other when a failure occurs, so as to ensure that the system based on the cloud storage architecture Reliable performance at runtime.

The data storage system for the safety production cloud service platform for industrial and mining enterprises is characterized in that: the two cloud storage management nodes can respectively manage three cloud storage spaces and multiple virtual devices, and the management includes Create, delete and configure three cloud storage spaces and multiple virtual devices for system backup, recovery and capacity expansion. Each cloud storage management node has a data directory. The data directory is used to record the relevant information of cloud storage space and virtual devices. The cloud storage management node finds the corresponding cloud storage space and virtual equipment; in addition, each cloud storage management node is provided with a unified application access entry, the application access entry is an application program interface, and the application program/service accesses the cloud storage space by calling the application access entry; the external application Programs/services are respectively connected to cloud storage management nodes a and b in the cloud storage architecture, and a unified application program access entry is set on the cloud storage management node, and the application program/service will indicate access or access through the application program access entry For the corresponding information in any cloud storage space, or to manage cloud storage space and virtual devices, the application/service connects to the cloud storage management node by calling the application access portal, and then the application/service specifies the cloud to be accessed through the API Storage space, file name, offset, access operation, the cloud storage management node assigns the final operation to one or more specific physical storage devices to complete the access operation according to the information passed in by these APIs and the internal data dictionary. Finally, the access result is returned through the cloud storage management node.

Description of drawings

Figure 1 The safety production cloud service platform system of industrial and mining enterprises

Figure 2 Integrated real-time data processing system for industrial and mining enterprises

Figure 3 wireless data acquisition equipment cluster

Figure 4 Structure diagram of wireless data acquisition equipment

Figure 5 Cloud storage system structure diagram

Figure 6 Cloud storage data center structure diagram

Figure 7 Real-time monitoring model of equipment resources

Figure 8 Equipment performance real-time monitoring model

Figure 9 SVM training and prediction flow chart

Figure 10 Immune evolution algorithm flow

Figure 11 Association rule analysis process

Figure 12 Petri net mining model

detailed description

Figure 1 shows a cloud service platform for safety production of industrial and mining enterprises, which integrates the functions of real-time data processing, safety production management business collaboration and system management that support the safety production service cloud, and realizes real-time monitoring and monitoring of enterprise production safety. early warning.

Specifically, the safety production cloud service platform system for industrial and mining enterprises includes the safety cloud service platform portal subsystem, the system management and related tool set development subsystem, the application service layer subsystem, the virtual resource layer subsystem, and the platform Service support layer subsystem, access and adaptation layer subsystem, security service resource subsystem, infrastructure service support layer subsystem.

The portal subsystem of the security cloud service platform is responsible for external information query and management based on web2.0 technology, including a government monitoring platform, an enterprise application platform, and an industry monitoring platform. The government monitoring platform is responsible for the government's real-time monitoring of enterprise operation equipment data and fiscal and taxation data, the enterprise application system is responsible for providing value-added services and application inquiries, and the industry supervision platform is used for product quality monitoring and feedback.

The system management and related tool set development subsystem includes development, deployment, monitoring, security management, log management and configuration of cloud service tools.

The application service layer includes a massive data processing system and a business collaborative operating system, wherein the massive data processing system is responsible for data collection, data integration, data management and provides standardized service interfaces to provide data streams for the security cloud service platform portal , and the business collaborative operating system is responsible for constructing a mathematical model for collaborative tasks and cross-domain tasks, specifically collecting business data, integrating service content provided by the cloud platform with collected business data, and predicting the development of subsequent businesses. Cross-domain tasks specifically include the management and monitoring of different business areas, and coordinate tasks between different departments of the enterprise.

The virtual resource layer mainly provides cloud servers with virtualized digital resources, specifically including knowledge service resource pools, production service resource pools, and data information resource pools. The virtual resource layer includes these original knowledge services that can be provided 1. The data information resources of the production service, through the virtualization technology, integrate this part of the data content into the application service layer, and can be called by the massive data processing unit and the business collaboration unit at any time.

The access and adaptation layer subsystem mainly obtains data resources from relevant standards and verification test systems, and also includes security service resources, specifically including production equipment data, standardization basic specifications, laws and regulations, safety management systems, accident sources Historical data, education and training knowledge, investment in safety production, organizational structure and responsibility, and hidden danger inspection information are adapted and accessed from third-party services.

The platform service support layer provides cloud service management and support engines, transaction collaboration logic engines, knowledge aggregation and classification engines, and cloud service management and support engines for the above-mentioned application service layer, virtual resource layer, and access adaptation layer services. Provide cloud service registration, release, cancellation, cloud service search, scheduling, combination, cloud service execution and monitoring for the application service layer; the transaction collaboration logic engine provides process management, cost accounting, and credit evaluation for business collaboration, and the The knowledge aggregation and classification engine provides industry multi-resource decentralized knowledge acquisition, industry knowledge modeling, and industry knowledge aggregation and classification for business collaboration, while other in the platform service support layer are responsible for operation management, operation and maintenance management, terminal software development, platform The module of the development tool serves the virtual resource layer and the access and adaptation layer, wherein the operation management is responsible for multi-tenant services, order management, delivery management, payment management, user management, credit management, and the operation and maintenance management is responsible for Security management, performance management and optimization, system configuration, massive data fault tolerance and reliability management, terminal software development, including sensor information fusion management, service resource image interface and universal human-computer interaction tools, and platform development tools , the platform service support layer is uniformly supported by the infrastructure service support layer through cloud computing, cloud network, and cloud storage.

The real-time data processing system of industrial and mining enterprises based on cloud computing, as shown in Figure 2, consists of five major parts: data acquisition equipment, data integration equipment, data management equipment, standardized service interface equipment, and high-speed and reliable optical fiber communication network, forming a cloud computing-based system. Real-time data integrated processing system for industrial and mining enterprises,

Data collection equipment: the network structure of the data integration system shown in Figure 2. The wireless sensor cluster is responsible for collecting data on the industrial and mining sites and transmitting the data to the industrial and mining data integration equipment. The data integration equipment is equipped with standard HTTP proxy module proxy The server and the cloud management server both use the ubuntu server blade board as the node to be managed, and a node manager is installed on each node server. The main function of this manager is to realize the management of the KVM virtual machine, including : (1) Receive the control command of the cloud manager to deploy, start and stop the KVM virtual machine; (2) Monitor the availability of the local KVM virtual machine, and try to start another same KVM when the local KVM virtual machine is unavailable due to an error Virtual machine instance; (3) Real-time monitoring of the usage status of each local KVM virtual machine resource, and sending the real-time resource usage status of the KVM virtual machine to the cloud manager; (4) Automatically execute the KVM virtual machine according to the real-time resource usage status of the virtual machine (5) receiving a virtual machine migration instruction from the cloud manager, and at the same time, multiple KVM virtual machines will be deployed on each node server, and a node server will be deployed in each virtual machine. All above-mentioned node servers share the same network data storage. The local storage of each node server is only used to cache the running data of each virtual machine itself, and the image files and embodiments of the virtual machine will be cached in the shared network data storage, which can more easily support high availability and virtualization Machine migration operation. In addition, the data collected by industrial and mining data will also be randomly distributed to the network storage under the control of cloud control servers and proxy servers, and the application software on each node server will complete business collaboration and cross-domain task collaboration. Figure 3 shows that a wireless sensor data acquisition cluster is provided, and the wireless sensor data acquisition device is connected with the industrial and mining data equipment through a data or analog interface, a wired or wireless interface according to the functional requirements of the application components in the system, and the acquisition frequency Set the highest frequency and precision required by the function for data sampling, the collected data is cached in its own memory, and the data is compressed into a unified format and transmitted to the remote monitoring center through the wireless transceiver unit. The wireless sensor devices can also communicate synchronously or asynchronously to form a data collection cluster. The working principle diagram of wireless data acquisition equipment shown in Fig. 4, comprises; Core processing unit, touch screen, camera, microphone, clock, memory, power management, interface circuit, sensor circuit are formed; Described core processing unit comprises embedded controller, Audio interface circuit, video interface circuit, touch screen interface circuit, multi-channel serial interface, high-speed USB interface, wireless communication module, debugging interface and sensor interface circuit; said touch screen interface circuit, memory, audio interface circuit, multi-channel serial interface , the wireless communication module, the debugging interface and the high-speed USB interface are respectively bidirectionally connected with the embedded controller; the touch screen is bidirectionally connected with the touch screen interface circuit; the microphone and the camera are respectively connected to the input ends of the audio interface circuit and the video interface circuit; The multi-channel serial interface and the high-speed USB interface are respectively connected to industrial and mining data equipment; the output terminal of the video interface circuit is connected to the corresponding input terminal of the embedded controller; the embedded controller is connected to the Internet; the power management device is used to provide electric energy, The power supply is in the form of a combination of photovoltaic solar panels and battery packs, which can not only ensure the effective and stable long-term operation of the equipment, but also avoid the laying of power lines; the charge and discharge controller composed of MC3063 chips is used to provide stable voltage, and the sensor unit Acceleration, pressure, temperature, light sensors and load signal sensors are installed in the center, the displacement calculation is realized through the acceleration signal and the load signal, and the environment is monitored through the pressure and temperature sensors. The core processing unit is used for the acquisition, operation and storage of sensor signals, and the wireless communication module realizes the reception and transmission of wireless signals; the embedded controller can use CC2530ZigBee chip to conveniently realize the transmission and reception of 2.4GHz ISM band signals based on ZigBee For example, when used on a drilling machine, the drilling machine includes a motor, a four-bar linkage, a beam, a bracket, and a pumping rod; the motor drives the beam to move through the four-bar linkage, and the beam drives and the pumping rod moves up and down. Data acquisition The acceleration signal sensor and the load signal sensor on the device can be arranged on the drawer rod, so as to collect the motion parameters of the drawer rod. In addition, the wireless data acquisition unit shown can also be set on industrial and mining electrical equipment for real-time video, audio, and digital monitoring of working parameters, obtaining working status, and uploading the collected data to the server through a multi-protocol gateway. The collected data is analyzed for the working condition of the corresponding equipment in order to obtain its operating status.

Data storage layer: There is a cloud data storage pool, which processes massive data from various data sources in industrial and mining enterprises, generates themed data corresponding to different businesses with a unified interface, and stores these themed data in distributed In the distributed file system, when there is a task request, according to different task requests, multi-node and multi-task parallel computing and analysis are performed on the data stored in the distributed file system, and the analysis results are correspondingly analyzed according to different applications. show. As shown in Figure 5, the cloud storage pool for data processing provided by the present invention includes at least one cloud storage management node, at least one cloud storage space and at least one virtual device, and the cloud storage management node, cloud storage space and virtual device constitute a private cloud . In the embodiment shown in FIG. 4 , the cloud storage architecture includes two cloud storage nodes, three cloud storage spaces and multiple virtual devices, and the three cloud storage spaces and multiple virtual devices constitute a private cloud. Cloud storage management node a and cloud storage management node b are connected to each other, the two cloud storage management nodes take over each other, the load can be balanced between the two cloud storage management nodes and when one of the cloud storage management nodes fails Takeover to ensure the reliable performance of the system based on the cloud storage architecture. Among them, two cloud storage management nodes can respectively manage three cloud storage spaces and multiple virtual devices, and its management includes creating, deleting and configuring three cloud storage spaces and multiple virtual devices for system backup , Restoration and expansion. Each cloud storage management node has a data directory. The data directory is used to record the relevant information of cloud storage space and virtual devices. The cloud storage management node finds the corresponding cloud storage space and virtual equipment. In addition, each cloud storage management node is provided with a unified application access entry, which is an application program interface, and the application program/service accesses the cloud storage space by calling the application access entry. Each virtual device in Figure 4 can be mapped as a raw device, memory area, file system or memory file system in the operating system, and virtual management of physical memory, built-in disk, and disk arrays of various interfaces and protocols. In addition, since multiple virtual devices c1, virtual devices c2, ..., virtual devices cn are virtual devices with the same characteristics, in this embodiment, virtual devices c1, virtual devices c2, ... ..., the virtual device cn can form a virtual device group b, and the management of multiple virtual devices is simplified through the virtual device group b. Multiple virtual devices may respectively correspond to a physical storage device or a storage space in a physical storage device. In this embodiment, virtual device a1, virtual device a2, ..., virtual device an are respectively connected to physical storage device a1, physical storage device a2, ..., physical storage device an, Therefore, the data information collected from the industrial and mining site is stored in the above physical storage device. External applications/services a and b are respectively connected to cloud storage management nodes a and b in the cloud storage architecture. Since a unified application access entry is set on the cloud storage management node, the application/service accesses the corresponding cloud storage space by calling the interface function of the application access entry. The application program/service will indicate to access or access the corresponding information in any cloud storage space through the application program access entry, or manage the cloud storage space and virtual devices. The application/service connects to the cloud storage management node by calling the application access portal, and then the application/service specifies the cloud storage space, file name, offset, access operation, etc. through the API, and the cloud storage management node according to these The information passed in by the API is combined with the internal data dictionary to assign the final operation to one or more specific physical storage devices to complete the access operation, and finally return the access result through the cloud storage management node.

Data computing layer: corresponding to the cloud data computing platform, the cloud data computing platform is used to call the real-time collected data stored in the cloud data storage pool to calculate its characteristics according to the business public relations of the industrial and mining system, and establish a general data computing and analysis model. For example, calculating equipment performance, load capacity, work efficiency, safety degree, environmental parameters and other values, the data calculation module of the virtual system on the node controller can be added and set flexibly, and the data analysis and calculation module can be updated to obtain corresponding calculation values. The structure of the data center is shown in Figure 6. Including: during the cloud control service period, the network adapter and proxy server are connected to the data network and the Internet to monitor and control multiple storage nodes and the virtual machines on them. The task control module of the main controller is responsible for the unified deployment of the downstream node controllers Management, including operations such as adding, deleting, and migrating any amount of data-readable physical drives and storage media of the control system, the management model is responsible for analyzing and processing the collected load and resource usage information, and then handing it over to the controller for control. Computing nodes contain any number of virtual machines. Each node contains a node controller responsible for the control of virtual machine resources inside the node. The collaborative management engine is responsible for resource allocation and synchronization management. Application resources run in virtual machines, such as resource monitors, Multiple performance monitors including performance monitor and early warning monitor. Taking the performance monitor as an example to introduce the data statistics process, as shown in Figure 7, the real-time load and resource monitoring model selects the preset device type according to the sampled data type, and then invokes the statistical model for the device to analyze and process the sampling performance Data, generate the ideal allocation data of equipment, and then perform model calculation through the controller combined with the actual sampling data, output the actual allocation data, judge the equipment resource usage and load level, and then guide the master control node to control the allocation of system equipment resources. As for the load monitoring module, as shown in Figure 8, during the operation of the system, according to the collected data information, the generated curve is monitored in real time to observe whether there are sudden changes in the curve or abnormal values that do not conform to the fitted curve. If it exists, it means that the application in the current data center has a peak time, which proves that the resources required by the system need to be greatly changed. At this time, it is necessary to quickly analyze the usage status of the system resources, and check whether the resource requirements can be met when the system resources reach the maximum capacity. Because it is an outlier point, the currently obtained parameters cannot represent the overall load, performance and relationship model between resources, but if it is not processed in time, it will have a great impact on the performance of the system, so it is necessary to analyze the outlier point in time deal with. If the resources in the system resource pool meet the current needs, they will be directly handed over to the cloud controller for processing to quickly resolve the current outliers. If the resources do not meet the current needs, the backup device resources will be started, and a simple linear regression model will be used to predict the next 5 Minute workload, simple linear regression model can effectively capture the change of workload over time, even more complex historical data can be easily inductively predicted its load. The predicted workload is used as input to the model to estimate the device resource requirements required by the existing workload and the achievable performance of the system. Many complex factors will affect the performance of the application, such as environmental parameters, changes in the number of operators, etc. At this time, KCCA algorithm and long-distance correlation algorithm can be used to realize multivariate statistical analysis and modeling, and simultaneously analyze the impact of multiple influencing factors on system performance. According to the influence of the future, the model parameters are adjusted in real time to generate ideal equipment resource allocation data. The calculated distribution data can be actively queried by the user or passively pushed to the user through the standardized query interface or communication interface on the master control node. Among them, for those data information that needs to be collected in real time, it is necessary to update the data in time to ensure the service quality of the data center.

Data access layer: The cloud service access module has a cloud service access interface, which is used to find the cloud server through the optical fiber network according to the trigger of the application component, and obtain the corresponding calculation value from the cloud data computing platform in real time. The interface services provided by the data access layer include cloud service access interfaces provided by various data servers.

The above four layers are connected sequentially through a reliable and high-speed optical switching communication network. The communication ring network is closely related to the power line. The high-speed communication optical fiber line is laid along the power line to all intelligent units, providing a powerful information high-speed communication channel for enterprises. Compared with the data collection in the traditional industrial and mining system, the data processing system of the present invention can expand the data storage and calculation processing functions of the cloud, as well as the specific logical judgment function based on the data model, and complete different functions by configuring different hardware devices or systems These devices or systems with different functions have their own independent data processing units. The traditional real-time data acquisition and processing unit is composed of multiple isolated systems, the data is collected repeatedly and the data is incomplete, and the measurement and control, measurement, protection, and safety automatic devices are realized locally by different hardware devices. The disadvantage is that data collection is repeated, data is difficult to share, and system application functions are difficult to coordinate effectively. The application functions of measurement and control, protection, metering, and safety automatic devices in the present invention are all realized through clustering of functional components, and through high-speed network communication technology, unified data processing, calculation, and sharing are truly realized, which is conducive to improving the reliability of the system , Reduce system installation costs and maintenance costs.

The above-mentioned mass information processing cloud platform manages and stores the data information accumulated in the actual production of safety production enterprises in a unified manner. In addition, in order to provide functions such as data-based prediction and abnormal detection in the production process, and realize the safe production of enterprises, managers also It is hoped that the abnormal situation in production can be discovered in time, the reason can be found out, and countermeasures can be put forward in time to keep the normal production. The judgment of the production situation is divided into normal and abnormal situations, so it can be classified as a classification problem, and the support vector machine with better classification effect is used to judge the abnormal production situation. By analyzing the index data of safe production in the database, select product quality, composition, and actual production rate as the supporting data for judging whether the production situation is abnormal, and perform boxplot analysis and correlation analysis on the supporting original data to obtain the mutual influence between the data . When selecting data samples, due to the large magnitude gap between variables, it is first necessary to standardize the variable data.

${V_i}^{\′}=\frac{V_i-\mathrm{\μ}}{\mathrm{\σ}},$ i=1,2,3,...,n

Among them, Vi is the original variable value, μ is the average value of the original variable value, and σ is the original variable standard deviation.

After standardized processing, the training and testing of the support vector machine decision model can be carried out, as shown in Figure 9, the training sample set data is used for model training, and the training algorithm adopts the SMO algorithm (SequentialMinimalOptimization, sequence minimum optimization), and its classification The support vector of the model, and the parameters of the decision function f(x) are calculated according to the support vector. The improved support vector machine obtained after training is not optimal, because the initial reference template and some parameter settings in the algorithm will affect the training results. A more optimal model can be obtained by selecting specific parameters in the layered kernel and selecting software metrics. Input the tree data structure corresponding to the software module that needs to be predicted into the trained improved support vector machine, and get the output between [-1, +1]. If the output is greater than 0, there is no abnormality in the production quality of the product; otherwise , the output is less than 0 means that the production situation of the product quality is abnormal.

Between the virtual resource layer and the access and adaptation layer described in Figure 1, it is necessary to extract and produce related standards, knowledge, data, etc. to help production and decision-making, which requires the cloud platform to be able to search for useful knowledge in the database . In view of the connotation and characteristics of knowledge, the purpose and requirements of knowledge transfer, we also introduce the immune operator (Immuneerator) into the standard evolutionary programming algorithm. Combining knowledge, knowledge transfer and immune theory, the immune planning algorithm based on knowledge transfer realizes knowledge reasoning and discovery. In the actual operation process, the immune algorithm is a global optimization algorithm developed on the basis of the genetic algorithm. Most of the problems that the genetic algorithm can solve, the immune algorithm can effectively solve and the efficiency is better than the genetic algorithm. Using the immune algorithm Good optimization ability can access all nodes in the virtual resource layer and wireless sensor network to complete data collection and select the path scheme with the smallest total energy consumption, and finally realize the reduction of energy consumption of wireless sensor networks, reduce network system delay, and improve virtual resource utilization. The problem of layer interaction efficiency. As shown in Figure 10, firstly, according to the optimization goals and conditions, the problem to be solved is specifically analyzed and decomposed, and the most basic feature information or feature set is extracted; secondly, the feature information is processed to extract its It is transformed into a scheme to solve the problem under the local environment or optimal constraints; finally, this scheme is transformed into an immune operator in an appropriate form and used to generate new individuals. Based on the process of knowledge and knowledge transfer, on the basis of rationally extracting immune vaccines, immune evolution can be realized through vaccination and immune selection, so as to effectively treat the prior knowledge of the problem and improve the fitness of individuals.

Taking the workflow of immune collaborative fault diagnosis software as an example, in the study of immunology, the phenomenon of mutual promotion and inhibition among various immune cells can be understood as a unique form of coevolution-immune coevolution, which can be used for reference in the mechanism of immune coevolution , according to the characteristics of the excitation system fault diagnosis problem and the cooperative diagnosis mode, a multi-diagnosis model cooperative evolution diagnosis strategy is proposed. Population adjustment mechanism, etc. Evolution adopts the basic immune algorithm, and the immune collaborative diagnosis strategy can be formally described as follows: ICED = (CPD, CPDN, CEDA, CPCM), where: CPD: immune cell diagnosis population, CPDN: immune cell diagnosis population number, Immune Collaborative Diagnosis Algorithm for Diagnosis Population,

$\mathrm{<span\; class="notranslate">\; CEDA</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{cccccc}{\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 1,1</span>}}& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 1,2</span>}}& <span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span>& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; CPN</span>}}& {\mathrm{<span\; class="notranslate">\; DA</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}& {\mathrm{<span\; class="notranslate">\; CN</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}\\ {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 2,1</span>}}& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 2,2</span>}}& <span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span>& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; CPN</span>}}& {\mathrm{<span\; class="notranslate">\; DA</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; CN</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>\\ <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>& <span\; class="notranslate">\; \&CenterDot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>\\ <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>& <span\; class="notranslate">\; \&Center\; Dot;</span>\\ {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; CPN</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 1</span>}}& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; CPN</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}}& <span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&CenterDot;</span>& {\mathrm{<span\; class="notranslate">\; CED</span>}}_{\mathrm{<span\; class="notranslate">\; CPN</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; CPN</span>}}& {\mathrm{<span\; class="notranslate">\; DA</span>}}_{\mathrm{<span\; class="notranslate">\; CPN</span>}}& {\mathrm{<span\; class="notranslate">\; CN</span>}}_{\mathrm{<span\; class="notranslate">\; CPN</span>}}\end{array}\right]$

Among them, CED _ij = {PCM _ij , CEDP _ij , EDV _ij }, i, j=1, 2, ..., CPDN: the cooperative mode of the i-th diagnostic cell population and the j-th diagnostic cell population, PCMij represents the synchronous task evaluation, Decide whether the task needs to be completed in cooperation with other agents, CEDPij represents the synchronous evolution operator, EDVij represents the synchronous evaluation operator; DA _i = {s _i , g _i , p _i , f _i , d _i }, i=1, 2, ..., CPN: the immune evolution algorithm of the i-th diagnostic cell population, si represents the selection strategy of the i-th diagnostic cell population, gi represents the evolution operator of the i-th diagnostic cell population, including cell cloning, cloning suppression, etc. , pi represents the execution probability of the evolution operator of the i-th diagnostic cell population, fi represents the affinity function of the i-th diagnostic cell population, di represents the concentration function of the i-th diagnostic cell population; CNi: the i-th diagnostic cell The number of individuals contained in the population; CPCM={CPO, CPD, CPA, CPE}: immune cell evolution population control mode, CPO represents the population size operator, CPD represents the population concentration adjustment, CPA represents the population evolution objective function; CPE represents the population evaluation method. After adopting the above description method, if: CPD(k)={CPD ₁ ^k , CPD ₂ ^k , ..., CPD _CPN ^k } represents the k-th generation immune cell population, then the co-evolution of the k-th generation immune cell population is expressed as: CEDA {CPD(k)}=CEDA{{CPD ₁ ^k , CPD ₂ ^k ,..., CPD _CPN ^k }}, then the evolution from generation k-1 to generation k in immune collaborative diagnosis can be expressed as: CPCM(k) : CEDA{CPD(k-1)}→CEDA{CPD(k)}, k=1, 2, . . . The abnormal diagnosis process is roughly divided into six stages: ① genI = 1, the fault diagnosis domain is set to equipment fault set F = {F1, ..., Fi, ..., FM}, and the diagnosis model set is FDM = {FDM1, . .., FDMi, ..., FDMM}, M is the number of equipment fault models, and FDMi is the i-th fault diagnosis model; ② For a certain diagnosis task FDM_TASK, initialize the fault diagnosis model group FDM, and assign each diagnosis model weight w= {w1,...,wi,...,wM}, wi represents the diagnostic importance of the diagnostic model FDMi in the diagnostic fault task FDM_TASK; ③ Each fault model FDMi gives its own diagnostic sub-conclusion FDR= for the diagnostic task FDM_TASK {FDR1,...,FDRi,...,FDRM}, calculate the individual FDMi affinity and concentration of each diagnostic model, evaluate the fault diagnosis effect of FDM_TASK, if the end condition of fault diagnosis is met, turn to ⑥; ④genI=genI+1 , for the fault diagnosis model population FDM, select a new generation population from the previous generation population based on affinity and concentration values; ⑤ Apply immune operators (cloning, mutation, suppression) to individuals in the population to obtain a new fault diagnosis model population FDM, and assign new diagnostic model weight w={w1,...,wi,...,wM}, turn to ③; ⑥ end of fault diagnosis. In the fault diagnosis process at a certain moment, the immune cooperative fault diagnosis software module 5 will not end the diagnosis process until it has completed all the above-mentioned diagnostic actions and obtained a relatively satisfactory diagnosis result.

In addition, for the knowledge aggregation and classification engine in the platform service support layer, with the production and construction of safety production enterprises, the number and types of data materials have increased accordingly, and the relationship between data has become increasingly complex. Traditional qualitative analysis of safety accidents It is no longer suitable for the needs of a large amount of complex data, and it is proposed to use the analysis tools of association rules to mine the characteristics and laws of safety accident data, and find the strong association rules between the types of accidents and the factors of "man-machine-environment-management". Early warning of accidents. Collect historical disaster data and design survey data related to excavation to obtain data related to construction accidents. Clean up the construction accident data to ensure the accuracy and specification of the data. The construction accident data records are incomplete, inconsistent, and have wrong information, etc. Therefore, in order to ensure the validity of the data in the subsequent analysis, this stage needs to clean up such data, mainly to solve the human errors in the establishment of data files, And some special values in the data file that have a great influence on the statistical analysis results establish a multidimensional data model. Figure 11 shows a data analysis method based on the analysis process of association rules. According to the characteristics of construction accidents and the causes of accidents, data attributes are set, and accident data and survey data are extracted from the database to form accident sample data. The data cube is established from the three dimensions of accident occurrence time and equipment operating parameters, and the data cube is operated by aggregate query and connection statement of SQL language to complete the search process of frequent predicate sets and strong rules. Generate frequent predicate sets. Specifically, it is necessary to meet the minimum support and minimum confidence. Set the minimum support count to 3, so as to determine the predicate set without concept hierarchical dimension. The minimum confidence can be expressed by the conditional probability represented by the minimum support count of the itemset, so that using An improved classic Apriori algorithm produces frequent predicate sets. Generate strong association rules. Further, by describing the practical meaning of the association rules that meet the minimum support threshold and minimum confidence threshold, such as the time of the month and the operating parameters of the equipment, the association prediction of the possible accident level will help the early warning of industrial and mining accidents and assist enterprises. Scientific decision-making. In the specific mining process, each threshold can be set jointly by staff and domain experts. That is to say, whether the rules are correct and applicable depends on the setting of boundary conditions.

In view of the characteristics of frequent interactions and close collaboration in safety production management of industrial and mining enterprises, sort out the business types of enterprises, and realize safety management collaboration that supports business collaboration, including the following technologies:

Based on foreign identification standards, combined with my country's production technology level and the differences in the environment of each site, analytic hierarchy process and fuzzy mathematics can be used for comprehensive evaluation to determine whether it is a major hazard. The analysis and evaluation of major hazards includes the causes of various hazards, the probability of accidents, the scope of consequences, etc. The evaluation of major hazard sources is one of the key measures to control major industrial accidents. The evaluation of major hazards should be carried out from two aspects: inherent risk evaluation and actual risk evaluation. Inherent risk assessment mainly reflects the inherent characteristics of substances, the characteristics of the production process of hazardous substances and the internal and external environmental conditions of hazardous units. Realistic risk assessment considers various risk offset factors on the basis of the former, and reflects people's subjective initiative in controlling the occurrence of accidents and the expansion of accident consequences. The external environment offset factor is added to propose the following evaluation model:

$\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; B</span>}}_{\mathrm{<span\; class="notranslate">\; 111</span>}}<span\; class="notranslate">\; )</span>}_{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; W</span>}}_{\mathrm{<span\; class="notranslate">\; ij</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; B</span>}}_{\mathrm{<span\; class="notranslate">\; 112</span>}}<span\; class="notranslate">\; )</span>}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; \}</span>{\mathrm{<span\; class="notranslate">\; B</span>}}_{\mathrm{<span\; class="notranslate">\; 12</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; B</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>$

In the formula (B111)i——the material hazard evaluation value of the i-th substance; (B112)j——the process hazard evaluation value of the j-th substance; Wij——the j-th process and the i-th substance hazard B12—accident severity evaluation value; B21—offset factor of process equipment, container building structure; B22—offset factor of personnel quality; B23—offset factor of safety management.

Usually, the method of grading hazards is adopted, and a circular area is determined by the death radius, and this circular area is the range affected by the hazard. The scope of this circular area is applicable to the impact of the explosion, but it is not applicable to some other situations, such as involving gas leakage, which cannot be completely determined according to the death radius. Accurately judge and classify hazard sources, and then take corresponding preventive and first aid measures based on the evaluation results. The purpose of the rapid evaluation and grading of major hazard sources is to quickly evaluate and classify major hazard sources based on the data information collected in the major hazard source data, so as to facilitate the macro-level monitoring and management of major hazard sources by the competent government departments. The rapid evaluation method of hazards mainly evaluates the accident consequences that may be caused by major hazards. The main evaluation index is the predicted death radius of the accident, and the major hazards are classified according to the size of the death radius. First select the number of the hazard source, then search the relevant data in the database according to the number, analyze the data after obtaining it, and judge the number of substances contained in the hazard source data and whether it is toxic, so as to deal with the hazard in several ways source. The most important part is to calculate the death radius of the damage model of toxic substances, flammable and explosive substances, and finally classify the hazards by the maximum death radius.

In addition, enterprises can also strengthen production site safety management and production process control through the cloud platform. The hidden dangers in the production process and materials, equipment and facilities, equipment, passages, working environment, etc. should be analyzed and controlled. Implement operation permit management for high-risk operations such as hot-fire operations, operations in confined spaces, temporary power-use operations, and high-altitude operations, and strictly implement the approval procedures. As shown in Figure 12, based on the Petri process mining algorithm, the non-Petri net modeling process model is converted into a Petri net to solve the open problem of mining hidden tasks from logs, so that the mined Petri net contains A task node without a label. Analyze the logs of the safe production process, evaluate the data of the production process, and eliminate potential safety hazards in time. Based on the Petr algorithm, it can be used to support a workflow engine for composite services, including: an interface layer, a control layer, a physical layer, a storage layer, and a database for storing business process instances; after the workflow engine is deployed, through The interface layer receives business information sent by the business system or other interface systems, and the other interface systems include a resource management system, a service activation management system, and a billing and accounting system; the interface layer provides three types of interfaces, including API interface, Corba interface, and WebService interface facilitate the connection between the workflow engine and the business system; the business information includes the process instance of the business and the completion of the current link of the process instance; the control layer receives the business delivered by the interface layer After information, according to the process routing control method that supports the composite business, control the generation, scheduling, decomposition, merging, and end of the business process instance; and determine whether the business process instance will automatically flow to the next link, or need to wait in place; At the same time, the control layer calls the method provided by the entity layer to record the completion of the current link of the process instance and the flow result determined by the process routing control method, and the process result is returned through the interface layer; the method provided by the entity layer It is: the operation of adding, modifying, deleting and querying the management objects described inside the workflow engine. The management objects include: business process instance objects, process routing objects, current link objects of process instances, and process task objects ; Wherein, the process task object describes the specific tasks performed by the link of each process instance; the storage layer persistently saves the process instance information of the business through the database. A verification method for optimized scheduling can also be integrated into the process operation process, and the specific steps are:

(1) Scheduling plan generation: According to product sales plan, raw material procurement plan, equipment maintenance plan, product (intermediate, raw material) inventory information and equipment production capacity, production resource occupation, consumption, production cost and other production constraint information, generate a target model The mathematical model of optimal scheduling is established based on the data files, and the optimal scheduling scheme is calculated according to the optimization goals set by the user (maximum production capacity, maximum profit, or satisfying sales orders) and the model solver.

(2) Predict possible production abnormalities through calculation: collect the current product sales plan, raw material procurement plan (including arrival status), inventory information, equipment maintenance plan, energy supply plan, etc. in the enterprise resource planning system, and use mixed equipment constraints (including intermittent production equipment and continuous production equipment), capacity constraints, logistics balance constraints and energy constraints are tracked and calculated according to time to predict possible production abnormalities.

(3) Combining the optimized scheduling scheme (data obtained in step 1) with possible production abnormalities (data obtained in step 2), the production process is simulated by using graphic visualization ESCPetri-Nets technology. Taking the scheduling time as the axis, the dynamic curve shows the equipment working conditions and material balance (purchase-inventory-sales) of the production workshop, and simulates and verifies the production scheduling situation. Centering on the material, it dynamically displays the change trend of a certain material and the operating status of equipment related to the production and consumption of a certain material in a graphical way as the scheduling time changes.

Claims (1)

1. A data storage system for a safe production cloud service platform for industrial and mining enterprises, characterized in that it includes a safe cloud service platform portal subsystem, a system management and related tool set development subsystem, an application service layer subsystem, a virtual The resource layer subsystem, the platform service support layer subsystem, the access and adaptation layer subsystem, the security service resource subsystem, the infrastructure service support layer subsystem, the application service layer subsystem also includes a massive data processing system, composed of Data acquisition equipment, data integration equipment, data management equipment, standardized service interface equipment and high-speed and reliable optical fiber communication network are composed of five parts, which constitute a real-time data integrated processing system for industrial and mining enterprises based on cloud computing. Data storage layer, aiming at massive data from various data sources in industrial and mining enterprises, after processing it, generates theme data with unified interface corresponding to different businesses, and stores these theme data in the distributed file system. When a task is requested, according to different task requests, multi-node and multi-task parallel computing and analysis are performed on the data stored in the distributed file system, and the analysis results are displayed according to different applications; the data storage system Including two cloud storage management nodes a and b, three cloud storage spaces and multiple virtual devices, three cloud storage spaces and multiple virtual devices constitute a private cloud; cloud storage management node a and cloud storage management node b are connected to each other , the two cloud storage management nodes take over each other, the two cloud storage management nodes can balance the load and take over each other when one of the cloud storage management nodes fails, so as to ensure the system runtime based on the cloud storage architecture reliable performance; the two cloud storage management nodes can respectively manage three cloud storage spaces and multiple virtual devices, and the management includes creating, deleting and configuring three cloud storage spaces and multiple virtual devices, For system backup, recovery and expansion; each cloud storage management node has a data directory, which is used to record the relevant information of cloud storage space and virtual devices, and the cloud storage management node uses the relevant data in its internal data directory Find the corresponding cloud storage space and virtual device; in addition, each cloud storage management node is set with a unified application access entry, the application access entry is an application program interface, and the application program/service is accessed by calling the application program Entrance to access cloud storage space; external applications/services are respectively connected to cloud storage management nodes a and b in the cloud storage architecture, and a unified application access entrance is set on the cloud storage management node, and applications/services will pass through The application access entry indicates to access or access the corresponding information in any cloud storage space, or manage cloud storage space and virtual devices. The application/service connects to the cloud storage management node by calling the application access entry, and then the application The /service will specify the cloud storage space, file name, offset, and access operation through the API. The cloud storage management node will combine the information passed in by these APIs with the internal data dictionary to make The final operation is assigned to one or more specific physical storage devices to complete the access operation, and finally the access result is returned through the cloud storage management node; in addition, the data storage system of the safe production cloud service platform also provides an association-based The data analysis method of the rule analysis process, which sets the data attributes according to the characteristics of the construction accident and the cause of the accident, and extracts the accident data and survey data from the database to form the accident sample data. Create its data cube with each dimension, and use the aggregation query and connection statement of SQL language to operate the data cube, complete the search process of frequent predicate sets and strong rules, and generate frequent predicate sets, which need to meet the minimum support and minimum confidence. , set the minimum support count to 3, so as to determine the predicate set without concept hierarchical dimension, the minimum confidence is expressed by the conditional probability represented by the minimum support count of the itemset, so that the improved classic Apriori algorithm is used to generate frequent predicate sets, resulting in Strong association rules can further predict the level of possible accidents by describing the realistic meaning of association rules that meet the minimum support threshold and minimum confidence threshold.