CN113076698B - Dynamic multi-target collaborative optimization method and system based on workshop big data - Google Patents

Abstract

The invention relates to the field of sugarcane crushing process optimization control, and relates to a dynamic multi-objective collaborative optimization method based on workshop big data, including 1) collecting big data information of a system, and extracting flow characteristic information; A single analysis is performed to form an information database, and the information database includes the state variables and order parameters of the flow; 3) Taking a single flow parameter target as the optimization goal, it is coupled with two other different flow parameter systems to analyze and clarify the coupling variables and auxiliary parameters in the information database. Variables and flow-flow synergy law; 4) On the basis of realizing system optimization under the domination of order parameters by each flow, under the constraints of system coupling variables and auxiliary variables, the mixed chicken swarm algorithm is used to solve the objective function , the process parameters of the system are calculated, and the aspect of the present invention is in an optimal state under different production boundary conditions of sugarcane pressing, improving the pressing effect and pressing amount, and reducing energy consumption.

Description

Dynamic multi-objective collaborative optimization method and system based on workshop big data

technical field

The invention relates to the field of sugarcane pressing process design optimization control, and relates to a dynamic multi-objective collaborative optimization method based on workshop big data.

Background technique

Sugarcane juice extraction is the first link in the sugar making process. Whether the juice extraction process is smooth or not, whether the extraction rate, pressing amount and production energy consumption meet the standards will affect the smooth operation and economic benefits of the entire sugar factory. The process of extracting juice is a complex process with multiple factors, multiple constraints, multiple objectives, strong coupling, large nonlinearity and uncertainty. In the design of the press production system, it mainly relies on the method of empirical calculation and analysis, and the control of the production process pays more attention to the local control of several key process points. within a reasonable range to achieve balanced feeding and balanced pressing. Therefore, the process operation parameters such as the conveying link and the permeate water treatment link of each process in the pressing process have poor adaptability due to the change of the working conditions such as the pressing amount, the sugar content of the sugarcane material, and the crushing degree. The resulting dynamic "short board" effect is more prominent, and there are many problems and contradictions in its production efficiency, effect and energy consumption. At present, the production system of sugar mills in my country generally has a low degree of automation, high energy consumption, low safety rate, poor stability, and poor adaptability to fluctuations in crushing volume.

How to use big data thinking and combine artificial intelligence technology to mine valuable information from the massive data generated in the production process of the intelligent workshop to guide the optimal control of workshop operation is an urgent problem that needs to be solved in the production process. The sugarcane pressing process is a dynamic and continuous multi-process production system, which is abstracted into the interaction and mutual influence of material flow, energy flow and information flow. It can effectively solve the problems of low automation, high energy consumption, low safety rate, poor stability and poor adaptability to the fluctuation of the crushing production system of sugar mills in my country, and provide solutions for energy saving and emission reduction, high quality and high yield for sugar enterprises.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a dynamic multi-objective collaborative optimization method based on workshop big data, so as to break the limitation of the traditional pressing system on the effect and efficiency of pressing and extracting juice.

In order to achieve the above purpose, the dynamic multi-objective collaborative optimization method based on workshop big data proposed by the present invention acts on the automatic control system for workshop multi-factor, multi-constraint, multi-objective, strong coupling and large nonlinearity. The parameters of the process include the parameter system of material flow, energy flow and information flow,

The steps of the dynamic multi-objective collaborative optimization method based on workshop big data include:

1) Collect the big data information of the system, and combine the workshop big data resources including cleaning and denoising, integration, and conversion preprocessing operations to extract flow feature information;

2) Analyzing material flow, energy flow and information flow one by one to form an information database, the information database includes flow state variables and order parameters;

3) Take the single material-energy flow synergistic correlation, energy information flow synergistic correlation, and material information flow synergistic correlation as the optimization goals, and analyze the coupling and synergy with other two different flow parameter systems, and clarify the coupling variables, auxiliary variables and flow parameters in the information database. - the law of flow synergy;

4) On the basis that each flow achieves system optimization under the domination of order parameters, the coordinated action of multiple flows in step 3) is globally coordinated, and under the consistent constraints of system coupling variables and auxiliary variables, the hybrid chicken swarm algorithm is adopted. The objective function is solved and the process parameters of the system are calculated.

Further, the flow characteristic information in the step 1) includes the factor variables and the operation mechanism of the flow involved.

Further, the described mixed flock algorithm process includes:

1) The fitness formula is redefined, and the elite reverse learning mechanism is introduced into the initialization of the individual population;

2) The forward and reverse learning mechanism is introduced in the update iteration of the rooster subgroup, and the parental guidance mechanism and adaptive factor are introduced in the chick update iteration;

3) Introduce a niche mechanism to update and maintain external archives to ensure the diversity of populations.

Further, the flow parameter objectives are: material-energy flow cooperative association, energy information flow cooperative association, and material information flow cooperative association.

The invention also acts on the automatic control system of workshop operations through a dynamic multi-objective collaborative optimization system based on workshop big data, including:

The stream feature attribute extraction layer collects the big data information of the system, and extracts the stream feature information by combining the workshop big data resources including cleaning, denoising, integration, and conversion preprocessing operations, and the feature information includes the factor variables of the involved streams and operating mechanism;

The subsystem analysis layer analyzes the material flow, the energy flow and the information flow one by one to form an information database, and the information of the database includes the state variables and order parameters of the flow;

The local collaborative optimization layer takes the single material energy flow collaborative correlation, energy information flow collaborative correlation, and material information flow collaborative correlation as the optimization goals, analyzes the coupling and collaboration of two different flow systems in the subsystem analysis model, and clarifies the data in the database. Coupling variables, auxiliary variables and the law of flow-flow synergy;

In the system collaborative optimization layer, on the basis that each flow realizes the subsystem optimization under the domination of order parameters, the coordination of multiple flows in the local collaborative optimization layer is globally coordinated, and the consistent constraints of the system coupling variables and auxiliary variables Next, the mixed chicken flock algorithm is used to solve the objective function, and the process parameters of the system are calculated.

Further, the flow parameter objectives are: material-energy flow cooperative association, energy information flow cooperative association, and material information flow cooperative association.

beneficial effect

The method of the invention uses the method of material flow, energy flow and information flow coordination to optimize the overall coordination and optimization of the workshop operation system. method and entropy analysis and other methods to optimize the design model of the workshop operation process, and propose a coordinated solver of the mixed flock method to solve the model. Through the synergy of the three streams, the operating parameters of the juice extraction process are optimized, and the dynamic synergistic control of the material flow, energy flow and information flow in the juice extraction process is realized, so as to ensure that the system is in an optimal state under different production boundary conditions, and improve the extraction efficiency and efficiency. effect, reducing energy consumption.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can also be obtained according to the structures shown in these drawings without creative efforts.

Fig. 1 is an analysis diagram of the "three-stream" coupling relationship of the sugarcane pressing system according to the present invention.

FIG. 2 is a solution diagram of dynamic multi-objective collaborative optimization modeling based on big data driven according to the present invention.

FIG. 3 is a schematic diagram of the multi-objective collaborative optimization modeling of the pressing system according to the present invention.

FIG. 4 is a schematic diagram of the optimal solution strategy based on the mixed flock algorithm according to the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Example 1

As shown in Figure 1, the present invention takes the sugarcane pressing workshop as an example. The process of sugarcane pressing and extracting juice is a complex process with multiple factors, multiple constraints, multiple objectives, strong coupling, large nonlinearity and uncertainty. "Three-flow" coupling relationship, according to the basic characteristics of the substances in the sugarcane pressing system, the factors involved and the relationship of flow changes, the sugarcane pressing process is divided into three subsystems: material flow, energy flow and information flow. The material flow is processed. Main body, including sugar cane, permeate water, mixed juice, bagasse, etc.; energy flow is the driving force of the process, including thermal energy for heating permeate water, electrical energy for equipment operation, mechanical energy, etc.; information flow is material flow behavior, energy flow behavior and the outside world The reaction of environmental information and the sum of artificial regulation information, including bagasse light conversion, bagasse moisture content, conveyor belt speed, sugarcane flow rate, permeate water temperature, process index attribute, efficiency attribute, production boundary attribute, equipment load attribute, etc. Through the analysis of the coupling relationship in the sugarcane pressing process, the interaction between material flow, energy flow and information flow is preliminarily analyzed. The change of energy flow affects the conversion efficiency of material flow. For example, the energy flow of the press affects the flow of mixed juice, and the energy flow of the pump affects the flow of permeate juice/water. As the carrier of information flow, material flow and energy flow affect the information flow index. Information flow restricts the change of material flow and energy flow. For example, process index in information flow affects sugarcane flow and equipment energy flow.

As shown in Figures 2-3, the dynamic multi-objective collaborative optimization system based on workshop big data according to the present invention divides the sugarcane pressing system into a material flow subsystem, an energy flow subsystem and an information flow subsystem. According to the above-mentioned three-stream coupling relationship in the process of sugarcane pressing and juice extraction, the three-stream collaborative optimization modeling of sugarcane pressing and juice extraction system includes big data center, stream feature extraction layer, subsystem analysis layer, local collaborative optimization layer, system collaborative optimization layer, system Optimize the solution layer.

The big data center is used to collect historical data and real-time data such as the state parameters of the processed substances, the state parameters of energy generated during the pressing process, and the state parameters of the equipment in the process of sugarcane pressing and juice extraction. Big data information such as plan, raw material composition, energy data, equipment status, finished product quality, etc.

The flow feature extraction layer is used to collect the big data information of the system, combined with the workshop big data resources that have undergone preprocessing operations such as cleaning and denoising, integration, and transformation, to extract the flow feature information including the factor variables of the involved flow and the operation mechanism; The feature extraction layer provides reliable and reusable data resources for subsequent multi-level analysis. According to the basic characteristics of the material in the pressing system and the relationship between the flow changes, combined with the big data information of the workshop, the comprehensive application includes one or more of the following analysis methods: time asymptotic method, alternating frequency method, experience transformation method, deep learning method and principal component Analysis method and other factors involved in the analysis system.

The subsystem analysis layer uses the flow feature attribute extraction layer to analyze a single flow subsystem to form a database of certain information, the information database includes order parameters and state variables that play a dominant role in the flow, and the order parameters include material flow sequence parameters, energy flow sequence parameters, and information flow sequence parameters. Specifically, the system optimization goal is taken as the output, the flow feature extraction layer extracts each state variable as the input, the weight of each state parameter is determined based on the data-driven model, the key factors of system performance are obtained by multi-attribute decision-making, and the state parameter system of each flow subsystem is established. ;

The local collaborative optimization layer is used to take the material-energy flow collaborative correlation, energy information flow collaborative correlation, and material information flow collaborative correlation as the optimization goals, and use entropy analysis and mutual information methods in combination with the database to analyze the two different subsystems in the subsystem analysis layer. The flow system coupling and synergy are analyzed, and the auxiliary variables, coupling variables and flow-flow synergy laws related to each target of the database are clarified.

In the system collaborative optimization layer, the system optimization design of material flow, energy flow and information flow coordination is used as the optimization model of this layer, and the optimization objectives include sugarcane juice extraction prediction, squeezing amount prediction, energy consumption prediction, and collaborative parameters, including process Gradient properties; experience properties, equipment load properties, process specification properties, operating tolerance properties as constraints. Based on the optimization of subsystems under the domination of order parameters, each flow takes the target set as the output and the auxiliary variables and coupling variables in the corresponding flow characteristic attributes as the input. Based on the deep data-driven method, the local The multiple streams between the collaborative optimization layers cooperate with each other for global coordination, and a collaborative optimization model for the pressing process is established.

The system optimization solution layer uses the hybrid chicken flock algorithm to calculate the process parameters of the system for the system collaborative optimization layer, which acts on the automatic control system to ensure that the system is in an optimal state under different production boundary conditions, and improves the pressing effect and efficiency. Squeeze volume and reduce energy consumption.

As shown in Figure 4, the schematic diagram of the optimization solution strategy based on the hybrid chicken swarm algorithm, the system target set {f ₁ (X), ......, f _M (X)}, M is the number of target sets Enter the optimization solution layer, and randomly generate particles of multiple dimensions, each particle corresponds to the value of a set of operating parameters under the current operating conditions. When all particles are traversed and optimized, each particle is input into the system process model obtained by the collaborative optimization layer of the system, and the real-time system operation target is obtained through the deep data-driven learning algorithm. Among them, the operating goal is to reduce energy consumption while ensuring high extraction and high pressing. The specific optimization process is as follows:

1) Set the evolutionary algebra of the population, the algebra of hierarchical system update, the range of constraint variables, the capacity of external files, and the proportion of the three subgroups, introduce a reverse learning strategy, and find the reverse population of randomly generated particles, and find the reverse population between the initial population and the reverse population. From the population, select the population with better fitness to form the initial population NP. The fitness calculation includes:

2) According to the set parameters for the individuals of the initial population, the entire solution space is divided into three subgroups according to the order of fitness, which are respectively the rooster group NR, the hen group NH and the chick group NC, according to the corresponding update formula Update subgroups.

3) Determine the optimal individual and the worst individual according to the proposed fitness.

4) Introduce a forward and reverse learning mechanism in the rooster subgroup, that is, forward learning to the global optimal individual to speed up the convergence speed. There is a certain probability to jump out of the local optimal solution.

x ^t+1 _i =x ^t _i *(1+Randn(0,σ ² ))+w ₁ (x ^t _best -x ^t _i )

x ^t+1 _i =x ^t _i *(1+Randn(0,σ ² ))+w ₂ (x ^t _worst -x ^t _i )

In the formula, Randn(0,σ ² ) is a Gaussian distribution with mean 0 and standard deviation σ ² , x ^t _i is the position of the ith individual in the t-th iteration, x ^t+1 _i t+1 The position of the i-th individual at the second iteration, x ^t _best is the global optimal individual at the t-th iteration, x ^t _best is the global worst individual at the t-th iteration, w ₁ and w ₂ are the forward learning and The learning factor of reverse learning, f _i is the fitness of the i-th individual, f _k is the fitness of the k-th individual, k∈[1,NH].

5) The formula for updating the position of the hen subgroup.

x ^t+1 _i =x ^t _i +S ₁ *rand*(x ^t _r1 -x ^t _i )+S ₂ *rand*(x ^t _r2 -x ^t _i )

In the formula, x ^t _i is the position of the i-th individual at the t-th iteration, x ^t+1 _i is the position of the i-th individual at the t+1-th iteration, r ₁ is the rooster followed by the hen, r ₂ is a randomly selected rooster or hen from the whole flock, and r ₁ ≠r ₂ .

6) Introduce the parental guidance mechanism and adaptive factor into chick position update.

x ^t+1 _i =w*x ^t _i +λ ₁ *(x ^t _m -x ^t _i )+λ ₂ *(x ^t _r1 -x ^t _i )

为小鸡所跟随的公鸡个体。w为权重，λ₁、λ₂分别为向母鸡、公鸡学习因子。In the formula, x ^t _i is the position of the i-th individual at the t-th iteration, x ^t+1 _i is the position of the i-th individual at the t+1-th iteration, and x ^t _m is the position followed by the i-th individual. individual hens,

An individual rooster followed by a chick. w is the weight, λ ₁ and λ ₂ are learning factors from hens and roosters, respectively.

7) The obtained non-dominated solution set is stored in the external archive, and the external archive set is updated and maintained by introducing the niche sharing mechanism to ensure the diversity of the population.

8) When the number of iterations is reached, the traversal optimization ends, and the external file set is output. According to the actual production needs, the corresponding optimal solution is decided in the external archives.

The integration of the chicken swarm algorithm and the forward and reverse learning mechanism can not only process data independently of each other, but also coordinate with each other and work together; it can not only ensure the particle development and exploration of the solution space, but also jump out of the local optimum in time when the algorithm stagnates and matures prematurely. . The integration of niche technology ensures the effectiveness of the hybrid flock algorithm for solving the multi-peak function optimization problem.

Example 2

On the basis of above-mentioned embodiment 1, a kind of dynamic multi-objective collaborative optimization method based on workshop big data is provided, and its steps include:

1) Collect historical data and real-time data in the process of sugarcane pressing and juice extraction, including the state parameters of the processed substances, the state parameters of energy generated during the pressing process, and the state parameters of equipment, etc. Big data information such as ingredients, energy data, equipment status, and finished product quality. The big data information of the system is collected, combined with the workshop big data resources that have undergone preprocessing operations including cleaning, denoising, integration, and transformation, to extract the flow characteristic information including the factor variables of the flow involved and the operation mechanism.

2) Analyze material flow, energy flow and information flow one by one to form an information database, the information database includes flow state variables and order parameters; the order parameters include material flow order parameters, energy flow order parameters, and information flow order parameters . Specifically, the system optimization goal is taken as the output, the flow feature extraction layer extracts each state variable as the input, the weight of each state parameter is determined based on the data-driven model, the key factors of system performance are obtained by multi-attribute decision-making, and the state parameter system of each flow subsystem is established. .

3) Take the single material-energy flow synergistic correlation, energy information flow synergistic correlation, and material information flow synergistic correlation as the optimization goals, and analyze the coupling and synergy with other two different flow parameter systems, and clarify the coupling variables, auxiliary variables and flow parameters in the information database. - the law of flow synergy;

4) On the basis that each flow achieves system optimization under the domination of order parameters, the coordinated action of multiple flows in step 3) is globally coordinated, and under the consistent constraints of system coupling variables and auxiliary variables, the hybrid chicken swarm algorithm is adopted. The objective function is solved and the process parameters of the system are calculated.

The above descriptions are only the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Under the inventive concept of the present invention, the equivalent structural transformations made by the contents of the description and drawings of the present invention, or the direct/indirect application Other related technical fields are included in the scope of patent protection of the present invention.

Claims (3)

1. the dynamic multi-objective collaborative optimization method based on workshop big data, acting on the automatic control system of workshop operation, it is characterized in that, the parameter of described workshop operation process comprises the parameter system of material flow, energy flow and information flow, The steps of the dynamic multi-objective collaborative optimization method based on workshop big data include: 1) Collect the big data information of the system, and combine the workshop big data resources including cleaning and denoising, integration, and conversion preprocessing operations to extract flow feature information; 2) Analyze the material flow, energy flow and information flow one by one to form an information database, and the information database includes the state variables and order parameters of the flow; 3) Take the single material-energy flow synergistic association, energy information flow synergistic association, and material information flow synergistic association as the optimization goals, and analyze the coupling and synergy with other two different flow parameter systems, and clarify the coupling variables, auxiliary variables and flow parameters in the information database. - the law of flow synergy; 4) On the basis that each flow realizes the system optimization under the domination of the order parameter, the coordinated action of multiple flows in step 3) is globally coordinated, and under the consistent constraints of the system coupling variables and auxiliary variables, the hybrid chicken swarm algorithm Solve the objective function and calculate the process parameters of the system; Wherein, the described mixed chicken flock algorithm process includes: ① The fitness formula is redefined, and the elite reverse learning mechanism is introduced into the initialization of the individual population; ② The forward and reverse learning mechanism is introduced in the update iteration of the rooster subgroup, and the parental guidance mechanism and adaptive factor are introduced in the chick update iteration; ③ Introduce a niche mechanism to update and maintain external archives to ensure the diversity of populations. 2 . The dynamic multi-objective collaborative optimization method based on workshop big data according to claim 1 , wherein the flow characteristic information in the step 1) includes the factor variables and operation mechanism of the flow involved. 3 . 3. A dynamic multi-objective collaborative optimization system based on workshop big data, acting on the automatic control system of workshop operations, is characterized in that, including: The stream feature attribute extraction layer collects the big data information of the system, and extracts the stream feature information by combining the workshop big data resources including cleaning, denoising, integration, and conversion preprocessing operations, and the feature information includes the factor variables of the involved streams and operating mechanism; The subsystem analysis layer analyzes the material flow, the energy flow and the information flow one by one to form an information database, and the information of the database includes the state variables and order parameters of the flow; The local collaborative optimization layer takes the single material energy flow collaborative correlation, energy information flow collaborative correlation, and material information flow collaborative correlation as the optimization goals, and analyzes the coupling and coordination of two different flow parameter systems in the subsystem analysis layer, and defines the database. Coupling variables, auxiliary variables and flow-flow synergy law; In the system collaborative optimization layer, on the basis that each flow realizes the subsystem optimization under the domination of order parameters, the coordination of multiple flows in the local collaborative optimization layer is globally coordinated, and the consistent constraints of the system coupling variables and auxiliary variables Next, the mixed chicken flock algorithm is used to solve the objective function, and the process parameters of the system are calculated; Wherein, the described mixed chicken flock algorithm process includes: ① The fitness formula is redefined, and the elite reverse learning mechanism is introduced into the initialization of the individual population; ② The forward and reverse learning mechanism is introduced in the update iteration of the rooster subgroup, and the parental guidance mechanism and adaptive factor are introduced in the chick update iteration; ③ Introduce a niche mechanism to update and maintain external archives to ensure the diversity of populations.

Images