CN119918842A - Wind turbine full life cycle archive management system and management method - Google Patents

Abstract

The invention provides a full life cycle archive management system and a management method of a fan, wherein the system comprises an Internet of things data acquisition module, a full life cycle digital archive management module, an AI predictive maintenance analysis module, an intelligent scheduling and maintenance decision support module, a responsibility tracing and process optimizing module and a maintenance detail recording and optimizing process and spare part management, wherein the Internet of things data acquisition module is arranged on key components of a wind turbine generator to acquire vibration, temperature and stress data, the full life cycle digital archive management module is used for recording full process information from design to operation and maintenance of the fan based on a cloud database and a blockchain technology, the data integrity and traceability are guaranteed, the AI predictive maintenance analysis module is used for analyzing data by using deep learning and machine learning algorithms to predict faults, the intelligent scheduling and maintenance decision support module is used for generating an optimal maintenance plan by combining external factors. The management method comprises the steps of data acquisition, storage, analysis, maintenance planning, task execution and recording, effect evaluation, flow optimization and the like. Through the implementation of the cooperative work of the modules and the system method, the fan is accurately and efficiently managed in the full life cycle.

Description

Fan full life cycle archive management system and management method thereof

Technical Field

The invention relates to the field of wind power plant equipment management, in particular to a full life cycle file management system and a management method of a fan.

Background

With the worldwide growing demand for renewable energy, wind power generation is increasingly gaining importance as a clean, renewable energy source. However, as a core component of the wind power generation system, the wind power generation set faces many challenges, especially how to effectively manage and maintain the wind power generation set to ensure long-term stable operation and maximum power generation efficiency due to the complex structure and severe operation environment. The maintenance of the traditional wind turbine generator is mostly dependent on periodic inspection and repair after failure, and the passive maintenance mode is low in efficiency, potential failure can not be found timely, unplanned shutdown is often caused, maintenance cost is increased, and overall power generation benefit is reduced.

In recent years, with the development of advanced technologies such as internet of things (IoT), big data, artificial Intelligence (AI), blockchain and the like, new possibilities are provided for intelligent management and predictive maintenance of wind turbines. The Internet of things technology enables the real-time monitoring of the running state of the wind turbine generator to be realized, the big data and the AI algorithm are combined, early signs of equipment faults can be mined from mass data, predictive maintenance is achieved, and the application of the blockchain technology greatly enhances the transparency, safety and traceability of the data, and establishes a solid information foundation for the whole life cycle management of the wind turbine generator.

Although the above technologies have remarkable advantages in theory, in practice, how to systematically integrate these technologies and construct a comprehensive management system that can monitor in real time and intelligently predict, and simultaneously optimize decision and responsibility tracing is still a technical problem to be solved. Therefore, the intelligent fan management and predictive maintenance system which can cover the whole life cycle of the wind turbine generator set and integrates data acquisition, intelligent analysis, decision support, responsibility tracing and process optimization is developed, and the intelligent fan management and predictive maintenance system has important significance for improving the operation and maintenance efficiency, reducing the cost and guaranteeing the power supply stability of the wind power generation industry.

Disclosure of Invention

The invention mainly aims to provide a full life cycle file management system and a management method thereof for a fan, and solves the problems that the maintenance of the traditional wind turbine generator is mostly dependent on periodic inspection and repair after failure, the passive maintenance mode is low in efficiency, potential failure cannot be found in time, unplanned shutdown is often caused, the maintenance cost is increased, and the overall power generation benefit is reduced.

In order to solve the technical problems, the technical scheme adopted by the invention is that the system for managing the full life cycle archives of the fan comprises an Internet of things data acquisition module, a control module and a control module, wherein the Internet of things data acquisition module is configured on key components of a wind turbine generator and is used for collecting vibration, temperature and stress running state data in real time;

The full life cycle digital archive management module is established in a database system of the cloud, and all information from design, manufacture and installation to operation and maintenance is recorded by using a blockchain technology, so that the integrity and traceability of data are ensured;

The AI predictive maintenance analysis module is used for analyzing the received data based on a deep learning algorithm and a machine learning algorithm, identifying the state change trend and the abnormal mode of the equipment and predicting possible faults in the future;

The intelligent scheduling and maintenance decision support module is used for automatically generating an optimal maintenance plan by combining external factors such as geographic positions, weather conditions and power grid requirements and providing the optimal maintenance plan to a maintenance team;

And the responsibility tracing and process optimizing module records all maintenance activity details and responsibility people, and optimizes maintenance flow and spare part management strategy by utilizing data analysis.

In a preferred scheme, the acquisition module further comprises at least one type of sensor, wherein the sensor is selected from a vibration sensor, a temperature sensor and a stress monitor and is used for accurately capturing the change of physical parameters in the running process of the wind turbine.

In the preferred scheme, the full life cycle digital archive management module further comprises a distributed account book system based on a blockchain, ensures that the creation, updating and access of each equipment archive record are transparent and can not be tampered, and supports efficient inquiry and data sharing.

In a preferred scheme, the AI predictive maintenance analysis module adopts at least one machine learning model, including a support vector machine, a random forest or a neural network, to train the collected multidimensional data so as to realize early recognition and early warning of fault occurrence.

In the preferred scheme, the scheduling and maintenance decision support module can integrate an intelligent algorithm, can automatically adjust a maintenance plan, and considers maintenance cost, shutdown influence and spare part supply conditions so as to minimize operation and maintenance cost and maximize power generation efficiency.

In the preferred scheme, the responsibility tracing and process optimizing module is provided with a data mining function, analyzes the association between maintenance activities and faults, and provides basis for continuously improving maintenance strategies and improving operation and maintenance efficiency.

Preferably, a user-friendly mobile application is further included to enable the field engineer to receive maintenance task notifications, view detailed failure analysis reports, historical maintenance records, and operational guidelines.

Preferably, the computer program product comprises computer program code stored on a non-transitory computer readable storage medium, which when executed by a computer causes the computer to perform all the functions of the system described above.

In a preferred embodiment, the method comprises:

The method comprises the steps of S1, data acquisition, namely, arranging vibration sensors and stress monitors at the root parts and key node positions of fan blades of a wind turbine generator, arranging temperature sensors and vibration sensors at proper positions of stator windings, bearings and shells of a generator, arranging the temperature sensors, the vibration sensors and the stress monitors at different positions of a box body of a gear box, near each bearing seat and at key stress structure positions respectively, and acquiring vibration, temperature and stress physical parameter change data in the running process of the wind turbine generator according to set sampling frequency by each sensor;

S2, after the full life cycle digital archive management module receives data, the full life cycle digital archive management module classifies and stores the data in a cloud database according to data sources and types, and inputs design drawings, manufacturing process parameters and installation and debugging record non-real-time operation data of the fan;

Creating a unique hash value identifier for each equipment file record by using a distributed account book system based on a blockchain, recording a time stamp of data operation, operator information and contents before and after modification, ensuring that the data is created, updated and accessed transparently, cannot be tampered and can be traced, and only authorizing a user or a node to operate the data;

S3, data analysis and fault prediction, namely an AI predictive maintenance analysis module extracts wind turbine generator set operation data in a specific time period from a full life cycle digital archive management module and performs standardized processing on the wind turbine generator set operation data;

selecting a support vector machine, a random forest or a neural network machine learning model, training the model by using historical normal and fault data, adjusting model parameters to enable the model learning equipment to be normal and fault state data characteristics, inputting data acquired and preprocessed in real time into the trained model, calculating the deviation degree of the current state and the normal state of the equipment, and predicting the type and time range of the fault possibly occurring in the future;

s4, a maintenance plan making step, namely an intelligent scheduling and maintenance decision support module receives a fault prediction result, and interacts with a geographic information system, a weather monitoring system and a power grid scheduling system to acquire the geographic position of a fan, weather conditions and power grid demand information, and combines the internal maintenance resource conditions of an enterprise;

Aiming at minimizing operation and maintenance cost and maximizing power generation efficiency, meeting maintenance task technical requirements and safety specification constraint conditions, automatically generating an optimal maintenance plan by utilizing an integrated intelligent algorithm, wherein the optimal maintenance plan comprises maintenance task arrangement, operation flow and power generation plan adjustment suggestion;

S5, a maintenance task executing and recording step, wherein the intelligent scheduling and maintenance decision support module pushes a maintenance plan to a field engineer through a mobile application program, the field engineer receives a notification and checks a fault analysis report, a historical maintenance record and an operation guide, the field engineer carries tools and spare parts to the field of the fan according to requirements and carries out maintenance operation according to an operation flow, key operation steps and data are recorded;

S6, determining evaluation indexes such as running stability after equipment repair, power generation efficiency recovery condition and maintenance cost according to maintenance task targets, collecting running data before and after maintenance, maintenance activity record data and cost data from a full life cycle digital archive management module, analyzing by using a data analysis tool, mining maintenance activities and fault occurrence association, optimizing maintenance flows according to evaluation results and analysis conclusion, such as modifying maintenance operation specifications, adjusting maintenance cycle and optimizing spare part management strategy, and updating the optimized maintenance flows and strategy to the full life cycle digital archive management module.

In a preferred embodiment, the method comprises:

the specific algorithm method of the steps S1-S6 is as follows:

s11, a data acquisition step, wherein the acquired vibration data sequence is The temperature data sequence isThe stress data sequence isWhereinThe number of sampling points;

firstly, calculating a dynamic mean value of data, taking vibration data as an example, wherein the dynamic mean value formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic average of the individual sample points,Is the dynamic average of the previous sample point,Is the smoothing coefficient [ ]) The method has the effects that in the data acquisition process, the influence of the current data value can be reflected, the trend of the historical data can be considered, the mean value mutation caused by individual abnormal data points is avoided, and the mean value can better follow the dynamic change trend of the data;

Then calculating the dynamic standard deviation of the data, wherein the formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic standard deviation of the individual sampling points,The dynamic standard deviation can be used for measuring the discrete degree of the data and helping to judge the stability and abnormal fluctuation condition of the data;

S21, data storage and management steps, namely data storage algorithm, namely setting data blocks Comprising acquisition timeData sourceData typeSpecific data values;

Using hash functionsWherein I represents a connection operation,Is a custom hash function, for example, a polynomial-based hash function is employed:

;

Wherein the method comprises the steps of ,Is a randomly generated coefficient [ ]),Is the degree of the polynomial,The hash function is used for generating a unique identifier for the data block, ensuring the integrity and traceability of the data, and any modification of the data can cause huge change of the hash value so as to be easily detected;

S32, a data analysis and fault prediction step, namely adopting a mixed algorithm based on a multi-scale entropy MSE and a support vector machine SVM;

firstly, calculating multi-scale entropy and setting an original data sequence Coarse granulating to obtain different sizesThe following sequenceWhereinMultiscale entropyThe calculation steps of (a) are as follows:

For each scale The following sequenceCalculating the sample entropyWhereinIs the dimension of the embedding,Is a similar tolerance, the sample entropy formula is:

;

Wherein the method comprises the steps of Is satisfied withA kind of electronic deviceDimension vector pairRatio of number of (d) to total vector logarithm [ ]) The multi-scale entropy reflects the complexity and regularity of the data on different scales, the entropy value of the normally operated fan data on different scales has a certain characteristic range, and the entropy value can be changed in a fault state;

then, the multi-scale entropy under different scales is formed into feature vectors The fault classification and prediction are input into a support vector machine, and the support vector machine constructs an optimal classification hyperplaneWhereinIs the normal vector of the vector,Is a function of mapping input data into a high-dimensional space,Is an offset term, using kernel functionsConverting the nonlinear problem of the low-dimensional space into the linear problem of the high-dimensional space, thereby realizing the classification of fault types and predicting the time range of fault occurrence;

S41, maintenance planning step, maintenance planning algorithm, setting maintenance cost Including maintenance personnel labor costsReplacement cost of spare partsCost of equipment downtime lossLoss of power generation efficiency;

Cost of labor for maintenance personnelWhereinIs the number of maintenance personnel, and the number of maintenance personnel,Is the firstThe unit time wages of the individual maintenance personnel,Is its expected working time;

Replacement cost of spare parts WhereinIs a kind of spare parts, and is provided with a plurality of parts,Is the firstThe unit price of the spare parts of the species,Is the required number;

equipment downtime loss cost WhereinIs the rated power of the fan,It is the estimated downtime that is to be expected,Is the economic value of the unit electric quantity;

loss of power generation efficiency WhereinIt is the time of occurrence of the fault,It is the return to normal operation time that is to be performed,Is a function of actual generated power over time;

construction of a Multi-objective optimization function WhereinAndIs a weight coefficient [ ]) Solving the multi-objective optimization problem through intelligent optimization algorithms such as genetic algorithm and the like to obtain optimal maintenance planning parameters such as maintenance personnel arrangement, spare part allocation and maintenance time;

s51, maintenance task execution and recording steps, namely, maintenance task execution and recording algorithm, namely, recording each operation step in the process of maintenance task execution Execution time of (a)Operation resultAnd related data;

Constructing maintenance task execution vectorsStoring the maintenance result in a full life cycle digital archive management module so as to carry out maintenance effect evaluation and flow optimization subsequently;

S61, maintenance effect evaluation and flow optimization, wherein the maintenance effect evaluation algorithm is an operation stability index after equipment repair WhereinIs the number of sampling points within the evaluation period,Is the actual operational data of the device,The smaller the index, the more stable the equipment operation is illustrated;

Maintenance cost compliance index WhereinIs the maintenance cost which actually occurs,Is the cost of planning the maintenance and repair,The closer toThe better the maintenance cost control is illustrated;

Power generation efficiency recovery index WhereinIs the average power generated by the fan after maintenance,Is the average generated power before maintenance,Greater thanThe power generation efficiency is improved;

through comprehensive analysis of the indexes, an association rule Apriori mining algorithm is utilized to mine association relation between maintenance activities and fault occurrence;

Wherein association rules between certain specific combinations of maintenance operations and the probability of the device re-failing within a certain time are found: Wherein AndIs a maintenance operation, and is performed by,And optimizing the maintenance flow, such as adjusting the sequence of maintenance operations, increasing or decreasing the frequency of certain maintenance operations, according to the association rules and the evaluation results.

The invention provides a fan full life cycle archive management system and a management method thereof, and the beneficial effects of the invention are that:

1) The system can monitor the state of the wind turbine in real time through integrating an advanced Internet of things technology and an AI algorithm, accurately identify potential faults, remarkably improve the accuracy of prediction maintenance, reduce unexpected downtime and ensure the stable operation of a wind farm.

2) The intelligent scheduling and decision support module can dynamically adjust the maintenance plan according to actual conditions, optimize resource allocation, and combine spare part management and cost control strategies, so that the operation and maintenance cost is reduced to the greatest extent, and the overall economic benefit is improved.

3) The full life cycle digital archive management system constructed by using the block chain technology ensures the non-tamper property and high transparency of the data, enhances the credibility of equipment information, is convenient for tracking and auditing, and is beneficial to responsibility definition and compliance management.

Drawings

The invention is further illustrated by the following examples in conjunction with the accompanying drawings:

FIG. 1 is a flow chart of a method for managing a full life cycle file management system of a fan in accordance with the present invention;

FIG. 2 is a fault handling flow chart of the present invention;

fig. 3 is a maintenance flow chart of the present invention.

Detailed Description

Example 1

1-3, The system comprises an Internet of things data acquisition module, a control module and a control module, wherein the Internet of things data acquisition module is configured on key components of a wind turbine generator and is used for collecting vibration, temperature and stress running state data in real time;

The full life cycle digital archive management module is established in a database system of the cloud, and all information from design, manufacture and installation to operation and maintenance is recorded by using a blockchain technology, so that the integrity and traceability of data are ensured;

The AI predictive maintenance analysis module is used for analyzing the received data based on a deep learning algorithm and a machine learning algorithm, identifying the state change trend and the abnormal mode of the equipment and predicting possible faults in the future;

The intelligent scheduling and maintenance decision support module is used for automatically generating an optimal maintenance plan by combining external factors such as geographic positions, weather conditions and power grid requirements and providing the optimal maintenance plan to a maintenance team;

And the responsibility tracing and process optimizing module records all maintenance activity details and responsibility people, and optimizes maintenance flow and spare part management strategy by utilizing data analysis.

In a preferred scheme, the acquisition module further comprises at least one type of sensor, wherein the sensor is selected from a vibration sensor, a temperature sensor and a stress monitor and is used for accurately capturing the change of physical parameters in the running process of the wind turbine.

In the preferred scheme, the full life cycle digital archive management module further comprises a distributed account book system based on a blockchain, ensures that the creation, updating and access of each equipment archive record are transparent and can not be tampered, and supports efficient inquiry and data sharing.

In a preferred scheme, the AI predictive maintenance analysis module adopts at least one machine learning model, including a support vector machine, a random forest or a neural network, to train the collected multidimensional data so as to realize early recognition and early warning of fault occurrence.

In the preferred scheme, the scheduling and maintenance decision support module can integrate an intelligent algorithm, can automatically adjust a maintenance plan, and considers maintenance cost, shutdown influence and spare part supply conditions so as to minimize operation and maintenance cost and maximize power generation efficiency.

In the preferred scheme, the responsibility tracing and process optimizing module is provided with a data mining function, analyzes the association between maintenance activities and faults, and provides basis for continuously improving maintenance strategies and improving operation and maintenance efficiency.

Preferably, a user-friendly mobile application is further included to enable the field engineer to receive maintenance task notifications, view detailed failure analysis reports, historical maintenance records, and operational guidelines.

Preferably, the computer program product comprises computer program code stored on a non-transitory computer readable storage medium, which when executed by a computer causes the computer to perform all the functions of the system described above.

Example 2

Further described in connection with example 1, as shown in FIGS. 1-3, the method includes:

The method comprises the steps of S1, data acquisition, namely, arranging vibration sensors and stress monitors at the root parts and key node positions of fan blades of a wind turbine generator, arranging temperature sensors and vibration sensors at proper positions of stator windings, bearings and shells of a generator, arranging the temperature sensors, the vibration sensors and the stress monitors at different positions of a box body of a gear box, near each bearing seat and at key stress structure positions respectively, and acquiring vibration, temperature and stress physical parameter change data in the running process of the wind turbine generator according to set sampling frequency by each sensor;

S2, after the full life cycle digital archive management module receives data, the full life cycle digital archive management module classifies and stores the data in a cloud database according to data sources and types, and inputs design drawings, manufacturing process parameters and installation and debugging record non-real-time operation data of the fan;

Creating a unique hash value identifier for each equipment file record by using a distributed account book system based on a blockchain, recording a time stamp of data operation, operator information and contents before and after modification, ensuring that the data is created, updated and accessed transparently, cannot be tampered and can be traced, and only authorizing a user or a node to operate the data;

S3, data analysis and fault prediction, namely an AI predictive maintenance analysis module extracts wind turbine generator set operation data in a specific time period from a full life cycle digital archive management module and performs standardized processing on the wind turbine generator set operation data;

selecting a support vector machine, a random forest or a neural network machine learning model, training the model by using historical normal and fault data, adjusting model parameters to enable the model learning equipment to be normal and fault state data characteristics, inputting data acquired and preprocessed in real time into the trained model, calculating the deviation degree of the current state and the normal state of the equipment, and predicting the type and time range of the fault possibly occurring in the future;

s4, a maintenance plan making step, namely an intelligent scheduling and maintenance decision support module receives a fault prediction result, and interacts with a geographic information system, a weather monitoring system and a power grid scheduling system to acquire the geographic position of a fan, weather conditions and power grid demand information, and combines the internal maintenance resource conditions of an enterprise;

Aiming at minimizing operation and maintenance cost and maximizing power generation efficiency, meeting maintenance task technical requirements and safety specification constraint conditions, automatically generating an optimal maintenance plan by utilizing an integrated intelligent algorithm, wherein the optimal maintenance plan comprises maintenance task arrangement, operation flow and power generation plan adjustment suggestion;

S5, a maintenance task executing and recording step, wherein the intelligent scheduling and maintenance decision support module pushes a maintenance plan to a field engineer through a mobile application program, the field engineer receives a notification and checks a fault analysis report, a historical maintenance record and an operation guide, the field engineer carries tools and spare parts to the field of the fan according to requirements and carries out maintenance operation according to an operation flow, key operation steps and data are recorded;

S6, determining evaluation indexes such as running stability after equipment repair, power generation efficiency recovery condition and maintenance cost according to maintenance task targets, collecting running data before and after maintenance, maintenance activity record data and cost data from a full life cycle digital archive management module, analyzing by using a data analysis tool, mining maintenance activities and fault occurrence association, optimizing maintenance flows according to evaluation results and analysis conclusion, such as modifying maintenance operation specifications, adjusting maintenance cycle and optimizing spare part management strategy, and updating the optimized maintenance flows and strategy to the full life cycle digital archive management module.

In a preferred embodiment, the method comprises:

the specific algorithm method of the steps S1-S6 is as follows:

s11, a data acquisition step, wherein the acquired vibration data sequence is The temperature data sequence isThe stress data sequence isWhereinThe number of sampling points;

firstly, calculating a dynamic mean value of data, taking vibration data as an example, wherein the dynamic mean value formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic average of the individual sample points,Is the dynamic average of the previous sample point,Is the smoothing coefficient [ ]) The method has the effects that in the data acquisition process, the influence of the current data value can be reflected, the trend of the historical data can be considered, the mean value mutation caused by individual abnormal data points is avoided, and the mean value can better follow the dynamic change trend of the data;

Then calculating the dynamic standard deviation of the data, wherein the formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic standard deviation of the individual sampling points,The dynamic standard deviation can be used for measuring the discrete degree of the data and helping to judge the stability and abnormal fluctuation condition of the data;

S21, data storage and management steps, namely data storage algorithm, namely setting data blocks Comprising acquisition timeData sourceData typeSpecific data values;

Using hash functionsWherein I represents a connection operation,Is a custom hash function, for example, a polynomial-based hash function is employed:

;

Wherein the method comprises the steps of ,Is a randomly generated coefficient [ ]),Is the degree of the polynomial,The hash function is used for generating a unique identifier for the data block, ensuring the integrity and traceability of the data, and any modification of the data can cause huge change of the hash value so as to be easily detected;

S32, a data analysis and fault prediction step, namely adopting a mixed algorithm based on a multi-scale entropy MSE and a support vector machine SVM;

firstly, calculating multi-scale entropy and setting an original data sequence Coarse granulating to obtain different sizesThe following sequenceWhereinMultiscale entropyThe calculation steps of (a) are as follows:

For each scale The following sequenceCalculating the sample entropyWhereinIs the dimension of the embedding,Is a similar tolerance, the sample entropy formula is:

;

Wherein the method comprises the steps of Is satisfied withA kind of electronic deviceDimension vector pairRatio of number of (d) to total vector logarithm [ ]) The multi-scale entropy reflects the complexity and regularity of the data on different scales, the entropy value of the normally operated fan data on different scales has a certain characteristic range, and the entropy value can be changed in a fault state;

then, the multi-scale entropy under different scales is formed into feature vectors The fault classification and prediction are input into a support vector machine, and the support vector machine constructs an optimal classification hyperplaneWhereinIs the normal vector of the vector,Is a function of mapping input data into a high-dimensional space,Is an offset term, using kernel functionsConverting the nonlinear problem of the low-dimensional space into the linear problem of the high-dimensional space, thereby realizing the classification of fault types and predicting the time range of fault occurrence;

S41, maintenance planning step, maintenance planning algorithm, setting maintenance cost Including maintenance personnel labor costsReplacement cost of spare partsCost of equipment downtime lossLoss of power generation efficiency;

Cost of labor for maintenance personnelWhereinIs the number of maintenance personnel, and the number of maintenance personnel,Is the firstThe unit time wages of the individual maintenance personnel,Is its expected working time;

Replacement cost of spare parts WhereinIs a kind of spare parts, and is provided with a plurality of parts,Is the firstThe unit price of the spare parts of the species,Is the required number;

equipment downtime loss cost WhereinIs the rated power of the fan,It is the estimated downtime that is to be expected,Is the economic value of the unit electric quantity;

loss of power generation efficiency WhereinIt is the time of occurrence of the fault,It is the return to normal operation time that is to be performed,Is a function of actual generated power over time;

construction of a Multi-objective optimization function WhereinAndIs a weight coefficient [ ]) Solving the multi-objective optimization problem through intelligent optimization algorithms such as genetic algorithm and the like to obtain optimal maintenance planning parameters such as maintenance personnel arrangement, spare part allocation and maintenance time;

s51, maintenance task execution and recording steps, namely, maintenance task execution and recording algorithm, namely, recording each operation step in the process of maintenance task execution Execution time of (a)Operation resultAnd related data;

Constructing maintenance task execution vectorsStoring the maintenance result in a full life cycle digital archive management module so as to carry out maintenance effect evaluation and flow optimization subsequently;

S61, maintenance effect evaluation and flow optimization, wherein the maintenance effect evaluation algorithm is an operation stability index after equipment repair WhereinIs the number of sampling points within the evaluation period,Is the actual operational data of the device,The smaller the index, the more stable the equipment operation is illustrated;

Maintenance cost compliance index WhereinIs the maintenance cost which actually occurs,Is the cost of planning the maintenance and repair,The closer toThe better the maintenance cost control is illustrated;

Power generation efficiency recovery index WhereinIs the average power generated by the fan after maintenance,Is the average generated power before maintenance,Greater thanThe power generation efficiency is improved;

through comprehensive analysis of the indexes, an association rule Apriori mining algorithm is utilized to mine association relation between maintenance activities and fault occurrence;

Wherein association rules between certain specific combinations of maintenance operations and the probability of the device re-failing within a certain time are found: Wherein AndIs a maintenance operation, and is performed by,And optimizing the maintenance flow, such as adjusting the sequence of maintenance operations, increasing or decreasing the frequency of certain maintenance operations, according to the association rules and the evaluation results.

Example 3

Further described in connection with example 2, as shown in figures 1-3,

And (S1) data acquisition, namely installing vibration sensors and stress monitors at the root parts and key node positions of fan blades of the wind turbine generator, respectively installing temperature sensors and vibration sensors at proper positions of stator windings, bearings and shells of the generator, and respectively installing the temperature sensors, the vibration sensors and the stress monitors at different positions of a box body of the gearbox, near each bearing seat and at key stress structure positions. And each sensor collects vibration, temperature and stress physical parameter change data in the running process of the wind turbine generator according to a set sampling frequency (for example, data are collected every 5 milliseconds).

Taking vibration data as an example, the collected vibration data sequence is. Firstly, calculating a dynamic average value of data, wherein the dynamic average value formula is as followsWherein(This value may be adjusted based on actual data fluctuations). Assume thatWhen (when)In the time-course of which the first and second contact surfaces,And analogically calculating the dynamic average value of each sampling point. Then calculate the dynamic standard deviation, the formula isAssuming an initial standard deviationFrom the first few data points, the dynamic standard deviation of each sampling point is calculated by this formula. The data acquisition module of the Internet of things cleans the acquired original data (removes peak data with obvious abnormality, such as exceeding the average value)Data of (2) and format conversion, and transmitting the data to the full life cycle digital file management module in real time in a 4G wireless transmission mode.

And step two, data storage and management step2, namely after the full life cycle digital archive management module receives the data, storing the data in a cloud database according to data sources (fan blades, generators, gear boxes and the like) and types (vibration, temperature and stress) in a classified manner, and inputting non-real-time operation data such as design drawings, manufacturing process parameters, installation and debugging records and the like of the fan.

Setting data blockComprising acquisition timeData sourceData typeSpecific data values. Using hash functionsWherein(Here,,,,,These coefficients may be adjusted according to data security and uniqueness requirements). For example for a block of dataWherein(2024, 12, 15, 10 Hours, 30 minutes, 05 seconds) collection,,,ThenCalculating hash valueCreating a unique identification for the data ensures that the data is created, updated and accessed transparently, non-tamperable and traceable, and only authorized users or nodes can manipulate the data.

And thirdly, the data analysis and fault prediction step S3 is that the AI predictive maintenance analysis module extracts wind turbine running data of the past week from the full life cycle digital archive management module and performs standardization processing (normalizes the data to the interval of $ [0,1 ]).

A hybrid algorithm based on multi-scale entropy (MSE) and Support Vector Machines (SVM) is employed. Firstly, calculating multi-scale entropy and setting an original data sequenceCoarse granulating to obtain different sizesThe following sequenceWherein. For each scaleThe following sequenceCalculating the sample entropyWherein,(These parameters may be adjusted based on data characteristics and fault detection accuracy). The sample entropy formula is. For example for dimensionsIs a sequence of (2)Calculation ofAndThereby obtaining the sample entropy. Then, the multi-scale entropy under different scales is formed into feature vectors(Taken hereThe scale value may be selected according to the actual situation). Feature vectorInputting the fault classification and prediction information into a support vector machine, and constructing an optimal classification hyperplane by the support vector machineUsing radial basis functionsAs a kernel function in whichThe nonlinear problem of the low-dimensional space is converted into the linear problem of the high-dimensional space by using a kernel function, so that the classification of fault types (such as fan blade cracks, generator overheat, gear abrasion of a gear box and the like) and the time range for predicting the occurrence of faults (such as the possibility of faults in the future 48 hours) are realized.

The fourth maintenance plan making step (S4) is that the intelligent dispatching and maintenance decision support module receives the fault prediction result and interacts with the geographic information system, the weather monitoring system and the power grid dispatching system to obtain the geographic position (longitude) of the fanLatitude, latitude) Weather conditions (rain is present in three days in the future, wind speed is 3-5 m/s), power grid demand information (current in electricity consumption valley period, low demand on the power generation power of the fan), and internal maintenance resource conditions (3 maintenance personnel, skill level is high, medium and low respectively), 2 sets of fan blade spare parts are in spare part stock, 3 sets of generator bearing spare parts and the like) are combined.

Set maintenance costIncluding maintenance personnel labor costsReplacement cost of spare partsCost of equipment downtime lossLoss of power generation efficiency. Cost of labor for maintenance personnelSuppose that advanced maintenance personnel pay per hourMeta, medium level maintenance personnelMeta, low-grade maintenance personnelThe estimated working time is respectivelyThe time period of the time period,The time period of the time period,Hours, thenAnd (5) a meta. Replacement cost of spare parts。

Replacement of spare parts of fan blade1 Element needs to be replaced, and spare parts of generator bearing are pricedThe number of elements is 2, if the elements need to be replacedAnd (5) a meta. Equipment downtime loss costRated power of fanEstimated downtimeHourly, economic value of unit electric quantityMeta/kWh, thenAnd (5) a meta. Loss of power generation efficiencyAssume that the average power generated before maintenanceAverage power generation after maintenanceMaintenance time intervalFor 10 hours, then(Where the minus sign indicates an increase in power generation efficiency). Construction of a Multi-objective optimization function。

Is provided with,Then. Optimizing different maintenance planning schemes (e.g., different service person combinations, spare part usage schemes, maintenance schedules, etc.) by genetic algorithms to minimizeThe value, obtain the optimal maintenance plan, including arranging the high-grade and middle-grade maintenance personnel to begin the maintenance operation at 9 am on the next day, replace fan blade spare parts first, then check the condition of the generator bearing, predict the maintenance duration for 8 hours, and make corresponding power generation plan adjustment suggestion (for example, increase the power generation power of the peripheral fans appropriately during maintenance to maintain the power grid stable).

And step (S5) of executing and recording maintenance tasks, wherein the intelligent scheduling and maintenance decision support module pushes the maintenance plan to a field engineer through a mobile application program, and the field engineer receives the notification and checks a fault analysis report (the detailed description of the fault prediction is based on the abnormal multi-scale entropy of the vibration of the fan blade, the similarity reaches 80% compared with historical fault data), a historical maintenance record (the last replacement of the fan blade spare part is performed for 300 days before one year, and the operation is stable after maintenance), and an operation guide (detailed fan blade spare part replacement steps and safety precautions). The field engineer carries tools and spare parts to the fan field according to the requirements, carries out maintenance operation according to the operation flow, and records each operation stepExecution time of (a)Operation resultAnd related data. For example, operating procedure "demolish old blade", execution timeMinute, result of operationRelated data. Constructing maintenance task execution vectorsThe responsibility tracing and process optimizing module stores the responsibility tracing and process optimizing module into the full life cycle digital archive management module, and a field engineer feeds back maintenance results (fan blade replacement succeeds, equipment runs normally, and vibration, temperature and stress data are restored to normal ranges) through a mobile application program after finishing tasks.

Step (S6) of maintenance effect evaluation and flow optimization, namely determining an evaluation index according to a maintenance task target, and operating stability index after equipment repairAssume that one week after maintenance is collectedA number of vibration data points,For the mean value of vibration data in normal operation, the vibration data is calculated to be $0.03$(Smaller values indicate more stable operation). Maintenance cost compliance indexCost of maintenance actually occursElement, plan maintenance costMeta-then(Approaching 1 illustrates better cost control). Power generation efficiency recovery index,,Then(Greater than 1 indicates an increase in power generation efficiency).

And excavating and maintaining the association relation between the activities and the fault occurrence through an Apriori algorithm. For example, the association rule of "maintenance person skill level is unbalanced and spare part inventory is insufficient" with "equipment fails again within three months" was found with a confidence of 60%. According to the evaluation results and the association rules, the maintenance flow is optimized, such as maintenance personnel training plan adjustment, low skill personnel level improvement, spare part inventory management strategy optimization, minimum inventory quantity of common spare parts increase and the like, and the optimized maintenance flow and strategy are updated into the full life cycle digital archive management module, so that reference basis is provided for subsequent maintenance management work, and experience reference is provided for maintenance management of other wind turbines.

In practical application, firstly, sensors are installed on each key component of a fan according to a data acquisition step (S1) and data acquisition and transmission are carried out. The full life cycle digital archive management module (S2) receives data and stores management. The AI predictive maintenance analysis module (S3) periodically performs data analysis and fault prediction. When a fault is predicted, the intelligent scheduling and maintenance decision support module (S4) makes a maintenance plan and pushes the maintenance plan. The field engineer executes the task according to the maintenance task execution and recording step (S5) and feeds back the result. And finally, continuously improving the maintenance flow and strategy through the maintenance effect evaluation and flow optimization step (S6), thereby realizing the efficient management of the whole life cycle of the fan, improving the operation reliability of the fan, reducing the operation and maintenance cost and improving the power generation efficiency.

The above embodiments are only preferred embodiments of the present invention, and should not be construed as limiting the present invention, and the scope of the present invention should be defined by the claims, including the equivalents of the technical features in the claims. I.e., equivalent replacement modifications within the scope of this invention are also within the scope of the invention.

Claims (10)

1. The system is characterized by comprising an Internet of things data acquisition module, a wind turbine generator system management module and a wind turbine generator system management module, wherein the Internet of things data acquisition module is configured on key components of the wind turbine generator system and used for collecting vibration, temperature and stress running state data in real time;

The full life cycle digital archive management module is established in a database system of the cloud, and all information from design, manufacture and installation to operation and maintenance is recorded by using a blockchain technology, so that the integrity and traceability of data are ensured;

The AI predictive maintenance analysis module is used for analyzing the received data based on a deep learning algorithm and a machine learning algorithm, identifying the state change trend and the abnormal mode of the equipment and predicting possible faults in the future;

The intelligent scheduling and maintenance decision support module is used for automatically generating an optimal maintenance plan by combining external factors such as geographic positions, weather conditions and power grid requirements and providing the optimal maintenance plan to a maintenance team;

And the responsibility tracing and process optimizing module records all maintenance activity details and responsibility people, and optimizes maintenance flow and spare part management strategy by utilizing data analysis.

2. The system of claim 1, wherein the collection module further comprises at least one type of sensor selected from a vibration sensor, a temperature sensor, and a stress monitor for accurately capturing changes in physical parameters during operation of the wind turbine.

3. The system of claim 1, wherein the full life cycle digital archive management module further comprises a distributed ledger system based on blockchain, ensuring that each device archive record is transparent and non-tamperable to be created, updated and accessed, and supporting efficient querying and data sharing.

4. The system of claim 1, wherein the AI predictive maintenance analysis module is configured to employ at least one machine learning model, including a support vector machine, a random forest, or a neural network, to train the collected multidimensional data to achieve early detection and early warning of failure.

5. The system of claim 1, wherein the scheduling and maintenance decision support module is capable of integrating an intelligent algorithm, automatically adjusting a maintenance schedule, and taking into account maintenance costs, downtime impact and spare part supply conditions to minimize operation and maintenance costs and maximize power generation efficiency.

6. The system for managing full life cycle archives of a fan as set forth in claim 1, wherein the responsibility tracing and process optimizing module is configured with data mining functionality to analyze correlations between maintenance activities and occurrences of faults, providing basis for continuously improving maintenance policies and improving operational and maintenance efficiencies.

7. A blower full life cycle archive management system as claimed in claim 1, further comprising a user-friendly mobile application enabling a field engineer to receive maintenance task notifications, view detailed failure analysis reports, historical repair records, and operational guidelines.

8. A fan full lifecycle profile management system, as set forth in claim 1, wherein the computer program product comprises computer program code stored on a non-transitory computer readable storage medium that, when executed by a computer, causes the computer to perform all the functions of the system.

9. The method for managing a full life cycle archive management system of a blower according to any one of claims 1 to 8, wherein the method comprises:

The method comprises the steps of S1, data acquisition, namely, arranging vibration sensors and stress monitors at the root parts and key node positions of fan blades of a wind turbine generator, arranging temperature sensors and vibration sensors at proper positions of stator windings, bearings and shells of a generator, arranging the temperature sensors, the vibration sensors and the stress monitors at different positions of a box body of a gear box, near each bearing seat and at key stress structure positions respectively, and acquiring vibration, temperature and stress physical parameter change data in the running process of the wind turbine generator according to set sampling frequency by each sensor;

S2, after the full life cycle digital archive management module receives data, the full life cycle digital archive management module classifies and stores the data in a cloud database according to data sources and types, and inputs design drawings, manufacturing process parameters and installation and debugging record non-real-time operation data of the fan;

Creating a unique hash value identifier for each equipment file record by using a distributed account book system based on a blockchain, recording a time stamp of data operation, operator information and contents before and after modification, ensuring that the data is created, updated and accessed transparently, cannot be tampered and can be traced, and only authorizing a user or a node to operate the data;

S3, data analysis and fault prediction, namely an AI predictive maintenance analysis module extracts wind turbine generator set operation data in a specific time period from a full life cycle digital archive management module and performs standardized processing on the wind turbine generator set operation data;

selecting a support vector machine, a random forest or a neural network machine learning model, training the model by using historical normal and fault data, adjusting model parameters to enable the model learning equipment to be normal and fault state data characteristics, inputting data acquired and preprocessed in real time into the trained model, calculating the deviation degree of the current state and the normal state of the equipment, and predicting the type and time range of the fault possibly occurring in the future;

s4, a maintenance plan making step, namely an intelligent scheduling and maintenance decision support module receives a fault prediction result, and interacts with a geographic information system, a weather monitoring system and a power grid scheduling system to acquire the geographic position of a fan, weather conditions and power grid demand information, and combines the internal maintenance resource conditions of an enterprise;

Aiming at minimizing operation and maintenance cost and maximizing power generation efficiency, meeting maintenance task technical requirements and safety specification constraint conditions, automatically generating an optimal maintenance plan by utilizing an integrated intelligent algorithm, wherein the optimal maintenance plan comprises maintenance task arrangement, operation flow and power generation plan adjustment suggestion;

S5, a maintenance task executing and recording step, wherein the intelligent scheduling and maintenance decision support module pushes a maintenance plan to a field engineer through a mobile application program, the field engineer receives a notification and checks a fault analysis report, a historical maintenance record and an operation guide, the field engineer carries tools and spare parts to the field of the fan according to requirements and carries out maintenance operation according to an operation flow, key operation steps and data are recorded;

S6, determining evaluation indexes such as running stability after equipment repair, power generation efficiency recovery condition and maintenance cost according to maintenance task targets, collecting running data before and after maintenance, maintenance activity record data and cost data from a full life cycle digital archive management module, analyzing by using a data analysis tool, mining maintenance activities and fault occurrence association, optimizing maintenance flows according to evaluation results and analysis conclusion, such as modifying maintenance operation specifications, adjusting maintenance cycle and optimizing spare part management strategy, and updating the optimized maintenance flows and strategy to the full life cycle digital archive management module.

10. The method for managing a full life cycle archive management system of a blower according to claim 9, wherein the method comprises the steps of:

the specific algorithm method of the steps S1-S6 is as follows:

s11, a data acquisition step, wherein the acquired vibration data sequence is The temperature data sequence isThe stress data sequence isWhereinThe number of sampling points;

firstly, calculating a dynamic mean value of data, taking vibration data as an example, wherein the dynamic mean value formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic average of the individual sample points,Is the dynamic average of the previous sample point,Is the smoothing coefficient [ ]) The method has the effects that in the data acquisition process, the influence of the current data value can be reflected, the trend of the historical data can be considered, the mean value mutation caused by individual abnormal data points is avoided, and the mean value can better follow the dynamic change trend of the data;

Then calculating the dynamic standard deviation of the data, wherein the formula is as follows:

;

Wherein the method comprises the steps of Is the firstThe dynamic standard deviation of the individual sampling points,The dynamic standard deviation can be used for measuring the discrete degree of the data and helping to judge the stability and abnormal fluctuation condition of the data;

S21, data storage and management steps, namely data storage algorithm, namely setting data blocks Comprising acquisition timeData sourceData typeSpecific data values;

Using hash functionsWherein I represents a connection operation,Is a custom hash function, for example, a polynomial-based hash function is employed:

;

Wherein the method comprises the steps of ,Is a randomly generated coefficient [ ]),Is the degree of the polynomial,The hash function is used for generating a unique identifier for the data block, ensuring the integrity and traceability of the data, and any modification of the data can cause huge change of the hash value so as to be easily detected;

S32, a data analysis and fault prediction step, namely adopting a mixed algorithm based on a multi-scale entropy MSE and a support vector machine SVM;

firstly, calculating multi-scale entropy and setting an original data sequence Coarse granulating to obtain different sizesThe following sequenceWhereinMultiscale entropyThe calculation steps of (a) are as follows:

For each scale The following sequenceCalculating the sample entropyWhereinIs the dimension of the embedding,Is a similar tolerance, the sample entropy formula is:

;

Wherein the method comprises the steps of Is satisfied withA kind of electronic deviceDimension vector pairRatio of number of (d) to total vector logarithm [ ]) The multi-scale entropy reflects the complexity and regularity of the data on different scales, the entropy value of the normally operated fan data on different scales has a certain characteristic range, and the entropy value can be changed in a fault state;

then, the multi-scale entropy under different scales is formed into feature vectors The fault classification and prediction are input into a support vector machine, and the support vector machine constructs an optimal classification hyperplaneWhereinIs the normal vector of the vector,Is a function of mapping input data into a high-dimensional space,Is an offset term, using kernel functionsConverting the nonlinear problem of the low-dimensional space into the linear problem of the high-dimensional space, thereby realizing the classification of fault types and predicting the time range of fault occurrence;

S41, maintenance planning step, maintenance planning algorithm, setting maintenance cost Including maintenance personnel labor costsReplacement cost of spare partsCost of equipment downtime lossLoss of power generation efficiency;

Cost of labor for maintenance personnelWhereinIs the number of maintenance personnel, and the number of maintenance personnel,Is the firstThe unit time wages of the individual maintenance personnel,Is its expected working time;

Replacement cost of spare parts WhereinIs a kind of spare parts, and is provided with a plurality of parts,Is the firstThe unit price of the spare parts of the species,Is the required number;

equipment downtime loss cost WhereinIs the rated power of the fan,It is the estimated downtime that is to be expected,Is the economic value of the unit electric quantity;

loss of power generation efficiency WhereinIt is the time of occurrence of the fault,It is the return to normal operation time that is to be performed,Is a function of actual generated power over time;

construction of a Multi-objective optimization function WhereinAndIs a weight coefficient [ ]) Solving the multi-objective optimization problem through intelligent optimization algorithms such as genetic algorithm and the like to obtain optimal maintenance planning parameters such as maintenance personnel arrangement, spare part allocation and maintenance time;

s51, maintenance task execution and recording steps, namely, maintenance task execution and recording algorithm, namely, recording each operation step in the process of maintenance task execution Execution time of (a)Operation resultAnd related data;

Constructing maintenance task execution vectorsStoring the maintenance result in a full life cycle digital archive management module so as to carry out maintenance effect evaluation and flow optimization subsequently;

S61, maintenance effect evaluation and flow optimization, wherein the maintenance effect evaluation algorithm is an operation stability index after equipment repair WhereinIs the number of sampling points within the evaluation period,Is the actual operational data of the device,The smaller the index, the more stable the equipment operation is illustrated;

Maintenance cost compliance index WhereinIs the maintenance cost which actually occurs,Is the cost of planning the maintenance and repair,The closer toThe better the maintenance cost control is illustrated;

Power generation efficiency recovery index WhereinIs the average power generated by the fan after maintenance,Is the average generated power before maintenance,Greater thanThe power generation efficiency is improved;

through comprehensive analysis of the indexes, an association rule Apriori mining algorithm is utilized to mine association relation between maintenance activities and fault occurrence;

Wherein association rules between certain specific combinations of maintenance operations and the probability of the device re-failing within a certain time are found: Wherein AndIs a maintenance operation, and is performed by,And optimizing the maintenance flow, such as adjusting the sequence of maintenance operations, increasing or decreasing the frequency of certain maintenance operations, according to the association rules and the evaluation results.