Coking co-production multi-objective optimization method and system based on global optimization strategy

Abstract

The invention discloses a coking co-production multi-objective optimization method and system based on a global optimization strategy, and the method comprises the steps: firstly constructing a historical data set, and training a coking co-production prediction model; then, a global multi-objective optimization engine is constructed, the global multi-objective optimization engine predicts the coke yield and the methanol yield based on the trained coking co-production prediction model, the multi-objective optimization model performs multi-objective optimization by adopting a multi-objective optimization function with constraint conditions, and a Pareto optimal solution set is obtained; the Pareto optimal solution set corresponds to control parameters of all sections of coking co-production, the control parameters of all sections of coking co-production form an optimal production decision scheme, and a control system of all sections of coking co-production performs collaborative automatic production of multiple sections in coking co-production according to the optimal production decision scheme. According to the method, the problem that the overall benefit is limited due to local optimization in traditional coking co-production is solved, and the overall operation efficiency, the economic benefit and the environmental protection level of a coking co-production system are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coking and co-production, and in particular to a coking and co-production multi-objective optimization method and system based on a global optimization strategy. Background Art

[0002] As an important basic industry of heavy industry, coking enterprises have begun to build optimization systems to optimize process operations, production management and business management in recent years to cope with multiple challenges such as cost control, efficiency improvement, resource utilization and environmental compliance.

[0003] At the same time, large-scale coking enterprises are developing in the direction of group operation and lean management, and gradually forming a new production model of coking and co-production (such as Figure 1 Its production process is responsible for converting coal into high-value-added chemical products such as coke, coal gas and methanol in sequence.

[0004] In this context, the managers of coking enterprises are faced with the new dilemma of organizing collaborative production on the one hand, and the challenge of simultaneously and globally optimizing key production variables distributed in different work sections on the other.

[0005] Existing optimization strategies for coking automation production systems are based on local optimization strategies, focusing on variable optimization within a single process section while ignoring the dynamic coupling relationships between multiple processes. These strategies are not suitable for the global optimization scenarios of multi-process collaborative production. Furthermore, traditional optimization strategies often focus on a single optimization objective. However, in actual production, coking cogeneration is not only a global optimization problem involving the coordination of multiple processes, but also a multi-objective optimization problem that requires consideration of multiple different optimization objectives. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a coking co-production multi-objective optimization method and system based on a global optimization strategy, which solves the problem of limited overall benefits caused by local optimization in traditional coking co-production, and significantly improves the overall operating efficiency, economic benefits and environmental protection level of the coking co-production system.

[0007] The technical solution of the present invention is: A multi-objective optimization method for coking and integrated production based on a global optimization strategy specifically includes the following steps: (1) Collect historical production variable parameters and production indicators of all sections in the coking co-production, and then perform data preprocessing to obtain a historical data set; the production indicators include coke production, gas production and methanol production; (2) Construct a coking joint production prediction model, and use historical data sets to train the coking joint production prediction model; (3) Construct a global multi-objective optimization engine. The global multi-objective optimization engine includes a trained coking joint production prediction model and a multi-objective optimization model. The global multi-objective optimization engine predicts the coke production and methanol production based on the trained coking joint production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set. The Pareto optimal solution set corresponds to the control parameters of all sections of the coking joint production. The control parameters of all sections of the coking joint production constitute the optimal production decision plan. The control systems of all sections of the coking joint production perform collaborative automated production of multiple sections in the coking joint production according to the optimal production decision plan. The multi-objective optimization function with constraints is specifically shown in the following formula (1): (1); In formula (1), represents a multi-objective optimization function, i.e., simultaneously satisfying the maximization of coke production, maximization of methanol production, and minimum cost of coal blending raw materials; Represents the coke production per unit time, Represents the methanol production per unit time, Represents the cost of coal blending raw materials per unit of output; represents the constraints of the multi-objective optimization function, Represents the minimum quality requirements for coke, Represents the process quality requirements of coke; Represents the gas production per unit time, represents the function of gas production determining methanol production, Represents the flue side temperature of the coke oven section, Represents the coke side temperature of the flue in the coke oven section, Represents the gas collecting pipe pressure of the coke oven section, including the gas collecting pipe input pressure and the gas collecting pipe output pressure, Represents the coking time of the coke oven section, Represented by four variable parameters Function that determines gas production; represents the minimum value of the variable parameter, Represents the maximum value of the variable parameter, Represents four variable parameters The value of is constrained.

[0008] The global multi-objective optimization engine performs multi-objective dynamic optimization based on the real-time production data of the coking and co-production. The real-time production data of the coking and co-production includes multiple variable parameters. The characteristic values ​​of the multiple variable parameters are normalized and then data verification is performed. After the data verification, all the verified and retained variable parameters are uploaded to the global multi-objective optimization engine for multi-objective dynamic optimization.

[0009] The data verification specifically includes the following steps: S11. Set the weights of all variable parameters in real-time production data , and set activation thresholds for all variable parameters ; S12. Calculate the activation score of each variable parameter , see the following formula (2) for details: (2); In formula (2), represents the eigenvalue of the i-th variable parameter, Represents the eigenvalue of the normalized parameter of the i-th variable; S13, when , then the i-th variable parameter is screened and retained; otherwise, it is screened and eliminated; S14. Summarize all retained variable parameters and upload them to the global multi-objective optimization engine for multi-objective dynamic optimization.

[0010] The real-time production data of the coking and production system includes variable parameters of four sections, specifically: Coal blending section: coal moisture, coal volatile matter, coal sulfur, coal ash, coal cohesiveness index, coal loading quantity per hole; Coke oven section: coke oven gas temperature before preheating, coke oven gas temperature after preheating, gas collecting pipe input end pressure, gas collecting pipe output end pressure, flue machine side suction, flue coke side suction, flue machine side temperature, flue coke side temperature and coking time; Gas section: primary cooler resistance, average gas temperature after primary cooler, electric precipitator resistance, gas pressure after mist collector, average blower oil temperature, average blower oil pressure, average gas pressure after desulfurization tower, gas temperature after benzene washing tower, gas pressure after benzene washing tower, lean oil temperature of benzene washing tower; Methanol section: coal gas usage, R104 catalyst bed temperature, compressor low-pressure cylinder inlet gas flow, compressor circulation section inlet flow, converter inlet coke gas temperature, converter coke oven gas and outlet conversion gas pressure difference, converter outlet conversion gas temperature, RT synthesis tower inlet temperature, RP synthesis tower pressure difference.

[0011] The coking joint production prediction model includes a data preprocessing module, a PDM module and an FMM module; The data preprocessing module aligns the variable parameters of all sections of each data sample according to the sampling time axis to form a unified multi-variable input matrix, that is, each data sample is processed by the data preprocessing module and outputs a time series data tensor with multiple sections, multiple variable parameters and different time steps. , , represents the time dimension, i.e. the length of the sequence, Total number of representative sections, , Represents the total number of variable parameters, time series data tensor The elements in , Represents the time step Time, work section Variable parameters under The value of The PDM module includes an L-layer structure, the input of the first layer structure is a time series data tensor , the input of each layer is , Represents the current layer, Represents the output of the previous layer structure, The total number of representative sections, Representative The temporal characteristics of the work section, the output of each layer of the structure is , the output of the last layer of structure That is the output of the PDM module; The input of the FMM module is , the FMM module uses an independent predictor for each section Predict the historical characteristics of the section and obtain future predictions , predictor As a linear layer, directly from the length The historical features are mapped to a length of The future time series of The final prediction is formed by weighted summing of the prediction results of all sections: , the output That is, the future production index of coking co-production predicted by the coking co-production prediction model.

[0012] The processing process of each layer structure of the PDM module is as follows: a. Extract seasonal components and mix: Let’s start with the residual, as shown in the following formula (3): (3); In formula (3), represents the initial residual, represents the initial trend component; Then the residual is calculated by the minimum period Perform period averaging to obtain the seasonal component of the first layer of coarse period , see the following formula (4) for details: (4); In formula (4), , represents the number of periods of the first layer of coarse period; Then perform multi-scale seasonal stacking and define the period set: , for each coarse cycle , calculate the residual , see the following formula (5) for details: (5); In formula (5), represents the residual of the coarse cycle of the m-1th layer, represents the residual of the coarse cycle of the m-2th layer, represents the seasonal component of the coarse cycle at the m-1th level; Then calculate the seasonal component of the mth layer coarse cycle , see the following formula (6) for details: (6); In formula (6), , represents the number of periods of the m-th layer coarse period; Finally, calculate the PDM module Mixed seasonal components of layer structure output , see the following formula (7) for details: (7); In formula (7), Representative The layer structure is based on the learnable weight coefficient of the seasonal component of the m-th layer coarse cycle for the work section s and the variable parameter v, satisfying the normalization constraint: ; b. Extract trend components and mix them: First, initialize the trend component, as shown in the following formula (8): (8); Formula (8), Represents a moving average operation, with an initial window size of ; Then the H layer trend is refined layer by layer, as shown in the following formula (9): (9); In formula (9), , , represents the trend component of the hth layer, represents the trend component of the h-1th layer, Represents the window size of the hth layer; Represents the learnable weight coefficient; Finally, calculate the PDM module Mixed trend component of layer structure output , see the following formula (10) for details: (10); In formula (10), Represents the trend component of the Hth layer, as the first Mixed trend component of layer structure output ; c. Cross-section coupling: Considering the work sections as graph nodes, the set of work section nodes is , construct the adjacency matrix , the adjacency matrix Elements in It represents the influence intensity of the i-th section on the j-th section; Then, the graph convolution method is used to perform cross-section feature fusion, as shown in the following formula (11): (11); In formula (11), is the adjacency matrix with self-loops added, represents the identity matrix; represent The degree matrix of ; Represents the PDM module Output of layer structure; Represents the PDM module The weight matrix of the layer structure, Represents the PDM module The dimension size of the layer structure input time series feature, Represents the weight matrix The dimension of the output feature after transformation; represents a nonlinear activation function; Represents the PDM module Cross-section fusion results of layer structures; Finally, the PDM module Cross-section fusion results of layer structure With mixed seasonal components , mixed trend component Fusion is performed to obtain the PDM module Output of layer structure , see the following formula (12) for details: (12); In formula (12), Represents a splicing operation.

[0013] The multi-objective optimization model is solved using a non-dominated sorting genetic algorithm III. That is, the real-time production data of the coking co-production is embedded in a multi-objective optimization function with constraints using the non-dominated sorting genetic algorithm III and solved to obtain a Pareto optimal solution set, any one of which is the optimal solution, thus forming a set of optimal production decision plans.

[0014] A coking and production multi-objective optimization system based on a global optimization strategy, including a data acquisition and verification module, a global multi-objective optimization engine, a collaborative production decision module, and a multi-section collaborative control module; The data acquisition and verification module is used to collect real-time production data of coking and co-production, normalize the characteristic values ​​of multiple variable parameters in the real-time production data, perform data verification, and upload all verified and retained variable parameters to the global multi-objective optimization engine; The global multi-objective optimization engine includes a trained coking co-production prediction model and a multi-objective optimization model. The coke production and methanol production are predicted based on the trained coking co-production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set. The collaborative production decision module corresponds the Pareto optimal solution set to the control parameters of all sections of the coking and co-production, and the control parameters of all sections of the coking and co-production form the optimal production decision plan; The multi-section collaborative control module is used to carry out collaborative automated production of multiple sections in coking joint production according to the optimal production decision-making plan, and feed back real-time production data to the data acquisition and verification module.

[0015] Advantages of the present invention: (1) The present invention uploads the real-time production data of the coking co-production to the global multi-objective optimization engine for multi-objective dynamic optimization after data verification, thereby reducing the amount of data transmission, lowering the load of the global multi-objective optimization engine, and improving the efficiency of prediction and the overall efficiency of evolutionary calculation.

[0016] (2) The coking joint production prediction model constructed by the present invention is based on the PDM module and FMM module of the TimeMixer model. It uses multi-scale observation to decouple seasonal and trend changes, and integrates the advantages of multiple predictors to jointly predict the output and quality indicators of each section, thereby significantly improving the overall prediction accuracy and response speed.

[0017] (3) The PDM module of the present invention gradually accumulates and mixes the seasonal components of all variable parameters of all sections at different scales from fine to coarse, thereby strengthening the capture of the short-term local dynamics of each section; the PDM module transmits the global long-term trend layer by layer from coarse to fine, and integrates it into the fine-grained changes of each section and each variable, so that the historical information of multiple sections is deeply layered and mixed. Finally, in the cross-section coupling, each section is no longer isolated, and the dynamic coupling and transition effects between sections are automatically mined through parameter sharing and high-dimensional embedding.

[0018] (4) The global multi-objective optimization engine constructed by the present invention predicts the coke production and methanol production based on the trained coking co-production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set to solve the optimization problem of multiple conflicting objectives in coking co-production. Without reducing the quality of coke, the future production indicators of coking co-production (coke production and methanol production) are made as high as possible and the cost of coal blending raw materials is made as low as possible, so as to maximize the comprehensive benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a process flow chart of coking co-production.

[0020] Figure 2 It is a flow chart of the multi-objective optimization method for coking co-production of the present invention.

[0021] Figure 3 It is a flow chart of the prediction of the coking co-production prediction model of the present invention.

[0022] Figure 4 It is a processing flow chart of each layer structure of the PDM module of the present invention.

[0023] Figure 5 It is a structural block diagram of the coking co-production multi-objective optimization system of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] See Figure 2 A multi-objective optimization method for coking and integrated production based on a global optimization strategy specifically includes the following steps: (1) Collect historical production variable parameters and production indicators of all sections in the coking co-production, and then perform data preprocessing to obtain a historical data set; production indicators include coke production, gas production and methanol production; (2) Construct a coking joint production prediction model, and use historical data sets to train the coking joint production prediction model; See Figure 3 ,The coking co-production prediction model includes data pre-processing module, PDM module and FMM module; The data preprocessing module aligns the variable parameters of all sections of each data sample according to the sampling time axis to form a unified multi-variable input matrix. That is, each data sample is processed by the data preprocessing module and outputs a time series data tensor with multiple sections, multiple variable parameters and different time steps. , , represents the time dimension, i.e. the length of the sequence, Total number of representative sections, , Represents the total number of variable parameters, time series data tensor The elements in , Represents the time step Time, work section Variable parameters under The value of The PDM module consists of an L-layer structure, where the input of the first layer is a time series data tensor. , the input of each layer is , Represents the current layer, Represents the output of the previous layer structure, The total number of representative sections, Representative The temporal characteristics of the work section, the output of each layer of the structure is , the output of the last layer of structure That is the output of the PDM module; See Figure 4 , the processing process of each layer of the PDM module is: a. Extract seasonal components and mix: Let’s start with the residual, as shown in the following formula (3): (3); In formula (3), represents the initial residual, represents the initial trend component; Then the residual is calculated by the minimum period Perform period averaging to obtain the seasonal component of the first layer of coarse period , see the following formula (4) for details: (4); In formula (4), , represents the number of periods of the first layer of coarse period; Then perform multi-scale seasonal stacking and define the period set: , for each coarse cycle , calculate the residual , see the following formula (5) for details: (5); In formula (5), represents the residual of the coarse cycle of the m-1th layer, represents the residual of the coarse cycle of the m-2th layer, represents the seasonal component of the coarse cycle at the m-1th level; Then calculate the seasonal component of the mth layer coarse cycle , see the following formula (6) for details: (6); In formula (6), , represents the number of periods of the m-th layer coarse period; Finally, calculate the PDM module Mixed seasonal components of layer structure output , see the following formula (7) for details: (7); In formula (7), Representative The layer structure is based on the learnable weight coefficient of the seasonal component of the m-th layer coarse cycle for the work section s and the variable parameter v, satisfying the normalization constraint: ; b. Extract trend components and mix them: First, initialize the trend component, as shown in the following formula (8): (8); Formula (8), Represents a moving average operation, with an initial window size of ; Then the H layer trend is refined layer by layer, as shown in the following formula (9): (9); In formula (9), , , represents the trend component of the hth layer, represents the trend component of the h-1th layer, Represents the window size of the hth layer; Represents the learnable weight coefficient; Finally, calculate the PDM module Mixed trend component of layer structure output , see the following formula (10) for details: (10); In formula (10), Represents the trend component of the Hth layer, as the first Mixed trend component of layer structure output ; c. Cross-section coupling: Considering the work sections as graph nodes, the set of work section nodes is , construct the adjacency matrix , the adjacency matrix Elements in It represents the influence intensity of the i-th section on the j-th section; Then, the graph convolution method is used to perform cross-section feature fusion, as shown in the following formula (11): (11); In formula (11), is the adjacency matrix with self-loops added, represents the identity matrix; represent The degree matrix of ; Represents the PDM module Output of layer structure; Represents the PDM module The weight matrix of the layer structure, Represents the PDM module The dimension size of the layer structure input time series feature, Represents the weight matrix The dimension of the output feature after transformation; represents a nonlinear activation function; Represents the PDM module Cross-section fusion results of layer structures; Finally, the PDM module Cross-section fusion results of layer structure With mixed seasonal components , mixed trend component Fusion is performed to obtain the PDM module Output of layer structure , see the following formula (12) for details: (12); In formula (12), Represents a splicing operation; The input of the FMM module is , the FMM module uses an independent predictor for each section Predict the historical characteristics of the section and obtain future predictions , predictor As a linear layer, directly from the length The historical features are mapped to a length of The future time series of The final prediction is formed by weighted summing of the prediction results of all sections: , the output That is, the future production index of coking co-production predicted by the coking co-production prediction model; (3) Collect real-time production data of coking and production, including four sections (see Figure 1 ), specifically: Coal blending section: coal moisture, coal volatile matter, coal sulfur, coal ash, coal cohesiveness index, coal loading quantity per hole; Coke oven section: coke oven gas temperature before preheating, coke oven gas temperature after preheating, gas collecting pipe input end pressure, gas collecting pipe output end pressure, flue machine side suction, flue coke side suction, flue machine side temperature, flue coke side temperature and coking time; Gas section: primary cooler resistance, average gas temperature after primary cooler, electric precipitator resistance, gas pressure after mist collector, average blower oil temperature, average blower oil pressure, average gas pressure after desulfurization tower, gas temperature after benzene washing tower, gas pressure after benzene washing tower, lean oil temperature of benzene washing tower; Methanol section: gas usage, R104 catalyst bed temperature, compressor low-pressure cylinder inlet gas flow, compressor circulation section inlet flow, reformer inlet coke gas temperature, reformer coke oven gas and outlet reformed gas pressure difference, reformer outlet reformed gas temperature, RT synthesis tower inlet temperature, RP synthesis tower pressure difference; Then, after the characteristic values ​​of multiple variable parameters are normalized, data verification is performed. The data verification specifically includes the following steps: S11. Set the weights of all variable parameters in real-time production data , and set activation thresholds for all variable parameters ; S12. Calculate the activation score of each variable parameter , see the following formula (2) for details: (2); In formula (2), represents the eigenvalue of the i-th variable parameter, Represents the eigenvalue of the normalized parameter of the i-th variable; S13, when , then the i-th variable parameter is screened and retained; otherwise, it is screened and eliminated; S14, summarizing all retained variable parameters and uploading them to the global multi-objective optimization engine for multi-objective dynamic optimization; (3) Construct a global multi-objective optimization engine. The global multi-objective optimization engine includes a trained coking co-production prediction model and a multi-objective optimization model. The global multi-objective optimization engine predicts the coke production and methanol production based on the trained coking co-production prediction model. The multi-objective optimization model is solved by a non-dominated sorting genetic algorithm III. That is, the real-time production data of the coking co-production is embedded in a multi-objective optimization function with constraints and solved to obtain a Pareto optimal solution set. Any solution (corresponding to the control parameters of each section of the coking co-production) is the optimal solution, thus forming a set of optimal production decision plans. The control systems of all sections of the coking co-production perform collaborative automated production of multiple sections in the coking co-production according to the optimal production decision plan. The multi-objective optimization function with constraints is shown in the following formula (1): (1); In formula (1), represents a multi-objective optimization function, i.e., simultaneously satisfying the maximization of coke production, maximization of methanol production, and minimum cost of coal blending raw materials; Represents the coke production per unit time, Represents the methanol production per unit time, Represents the cost of coal blending raw materials per unit of output; represents the constraints of the multi-objective optimization function, Represents the minimum quality requirements for coke, Represents the process quality requirements of coke; Represents the gas production per unit time, represents the function of gas production determining methanol production, Represents the flue side temperature of the coke oven section, Represents the coke side temperature of the flue in the coke oven section, Represents the gas collecting pipe pressure of the coke oven section, including the gas collecting pipe input pressure and the gas collecting pipe output pressure, Represents the coking time of the coke oven section, Represented by four variable parameters Function that determines gas production; Represents the minimum value of the variable parameter (control parameter that affects the production of the work section), Represents the maximum value of the variable parameter (control parameter that affects the production of the work section), Represents four variable parameters The value of is constrained.

[0026] See Figure 5 A coking joint production multi-objective optimization system based on a global optimization strategy includes a data acquisition and verification module 1, a global multi-objective optimization engine 2, a collaborative production decision module 3 and a multi-section collaborative control module 4; The data acquisition and verification module 1 is used to collect the real-time production data of the coking co-production, normalize the characteristic values ​​of multiple variable parameters in the real-time production data, perform data verification, and upload all the verified and retained variable parameters to the global multi-objective optimization engine; The global multi-objective optimization engine 2 includes a trained coking co-production prediction model and a multi-objective optimization model. It predicts the coke production and methanol production based on the trained coking co-production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set. The collaborative production decision module 3 corresponds the Pareto optimal solution set to the control parameters of all sections of the coking and co-production, and the control parameters of all sections of the coking and co-production form the optimal production decision plan; The multi-section collaborative control module 4 is used to carry out collaborative automated production of multiple sections in the coking joint production according to the optimal production decision plan, and feed back real-time production data to the data acquisition and verification module.

[0027] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-objective optimization method for coking and integrated production based on a global optimization strategy, characterized by: The specific steps include: (1) Collect historical production variable parameters and production indicators of all sections in the coking co-production, and then perform data preprocessing to obtain a historical data set; the production indicators include coke production, gas production and methanol production; (2) Construct a coking joint production prediction model, and use historical data sets to train the coking joint production prediction model; (3) Construct a global multi-objective optimization engine. The global multi-objective optimization engine includes a trained coking joint production prediction model and a multi-objective optimization model. The global multi-objective optimization engine predicts the coke output and methanol output based on the trained coking joint production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set. The Pareto optimal solution set corresponds to the control parameters of all sections of the coking joint production. The control parameters of all sections of the coking joint production constitute the optimal production decision plan. The control systems of all sections of the coking joint production perform collaborative automated production of multiple sections in the coking joint production according to the optimal production decision plan.

2. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 1, characterized in that: The multi-objective optimization function with constraints is specifically shown in the following formula (1): （1）； In formula (1), represents a multi-objective optimization function, i.e., simultaneously satisfying the maximization of coke production, maximization of methanol production, and minimum cost of coal blending raw materials; Represents the coke production per unit time, Represents the methanol production per unit time, Represents the cost of coal blending raw materials per unit of output; represents the constraints of the multi-objective optimization function, Represents the minimum quality requirements for coke, Represents the process quality requirements of coke; Represents the gas production per unit time, represents the function of gas production determining methanol production, Represents the flue side temperature of the coke oven section, Represents the coke side temperature of the flue in the coke oven section, Represents the gas collecting pipe pressure of the coke oven section, including the gas collecting pipe input pressure and the gas collecting pipe output pressure, Represents the coking time of the coke oven section, Represented by four variable parameters Function that determines gas production; represents the minimum value of the variable parameter, Represents the maximum value of the variable parameter, Represents four variable parameters The value of is constrained.

3. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 2, characterized in that: The global multi-objective optimization engine performs multi-objective dynamic optimization based on the real-time production data of the coking and co-production. The real-time production data of the coking and co-production includes multiple variable parameters. The characteristic values ​​of the multiple variable parameters are normalized and then data verification is performed. After the data verification, all the verified and retained variable parameters are uploaded to the global multi-objective optimization engine for multi-objective dynamic optimization.

4. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 3 is characterized in that: The data verification specifically includes the following steps: S11. Set the weights of all variable parameters in real-time production data , and set activation thresholds for all variable parameters ; S12. Calculate the activation score of each variable parameter , see the following formula (2) for details: （2）； In formula (2), represents the eigenvalue of the i-th variable parameter, Represents the eigenvalue of the normalized parameter of the i-th variable; S13, when , then the i-th variable parameter is screened and retained; Otherwise, it will be screened out; S14. Summarize all retained variable parameters and upload them to the global multi-objective optimization engine for multi-objective dynamic optimization.

5. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 3 is characterized in that: The real-time production data of the coking and production system includes variable parameters of four sections, specifically: Coal blending section: coal moisture, coal volatile matter, coal sulfur, coal ash, coal cohesiveness index, coal loading quantity per hole; Coke oven section: coke oven gas temperature before preheating, coke oven gas temperature after preheating, gas collecting pipe input end pressure, gas collecting pipe output end pressure, flue machine side suction, flue coke side suction, flue machine side temperature, flue coke side temperature and coking time; Gas section: primary cooler resistance, average gas temperature after primary cooler, electric precipitator resistance, gas pressure after mist collector, average blower oil temperature, average blower oil pressure, average gas pressure after desulfurization tower, gas temperature after benzene washing tower, gas pressure after benzene washing tower, lean oil temperature of benzene washing tower; Methanol section: coal gas usage, R104 catalyst bed temperature, compressor low-pressure cylinder inlet gas flow, compressor circulation section inlet flow, converter inlet coke gas temperature, converter coke oven gas and outlet conversion gas pressure difference, converter outlet conversion gas temperature, RT synthesis tower inlet temperature, RP synthesis tower pressure difference.

6. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 1, characterized in that: The coking joint production prediction model includes a data preprocessing module, a PDM module and an FMM module; The data preprocessing module aligns the variable parameters of all sections of each data sample according to the sampling time axis to form a unified multi-variable input matrix, that is, each data sample is processed by the data preprocessing module and outputs a time series data tensor with multiple sections, multiple variable parameters and different time steps. , , represents the time dimension, i.e. the length of the sequence, Total number of representative sections, , Represents the total number of variable parameters, time series data tensor The elements in , Represents the time step Time, work section Variable parameters under The value of The PDM module includes an L-layer structure, the input of the first layer structure is a time series data tensor , the input of each layer is , Represents the current layer, Represents the output of the previous layer structure, The total number of representative sections, Representative The temporal characteristics of the work section, the output of each layer of the structure is , the output of the last layer of structure That is the output of the PDM module; The input of the FMM module is , the FMM module uses an independent predictor for each section Predict the historical characteristics of the section and obtain future predictions , predictor As a linear layer, directly from the length The historical features are mapped to a length of The future time series of The final prediction is formed by weighted summing of the prediction results of all sections: , the output That is, the future production index of coking co-production predicted by the coking co-production prediction model.

7. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 6, characterized in that: The processing process of each layer structure of the PDM module is as follows: a. Extract seasonal components and mix: Let’s start with the residual, as shown in the following formula (3): （3）； In formula (3), represents the initial residual, represents the initial trend component; Then the residual is calculated by the minimum period Perform period averaging to obtain the seasonal component of the first layer of coarse period , see the following formula (4) for details: （4）； In formula (4), , represents the number of periods of the first layer of coarse period; Then perform multi-scale seasonal stacking and define the period set: , for each coarse cycle , calculate the residual , see the following formula (5) for details: （5）； In formula (5), represents the residual of the coarse cycle of the m-1th layer, represents the residual of the coarse cycle of the m-2th layer, represents the seasonal component of the coarse cycle at the m-1th level; Then calculate the seasonal component of the mth layer coarse cycle , see the following formula (6) for details: （6）； In formula (6), , represents the number of periods of the m-th layer coarse period; Finally, calculate the PDM module Mixed seasonal components of layer structure output , see the following formula (7) for details: （7）； In formula (7), Representative The layer structure is based on the learnable weight coefficient of the seasonal component of the m-th layer coarse cycle for the work section s and the variable parameter v, satisfying the normalization constraint: ; b. Extract trend components and mix them: First, initialize the trend component, as shown in the following formula (8): （8）； Formula (8), Represents a moving average operation, with an initial window size of ; Then the H layer trend is refined layer by layer, as shown in the following formula (9): （9）； In formula (9), , , represents the trend component of the hth layer, represents the trend component of the h-1th layer, Represents the window size of the hth layer; Represents the learnable weight coefficient; Finally, calculate the PDM module Mixed trend component of layer structure output , see the following formula (10) for details: （10）； In formula (10), Represents the trend component of the Hth layer, as the first Mixed trend component of layer structure output ; c. Cross-section coupling: Considering the work sections as graph nodes, the set of work section nodes is , construct the adjacency matrix , the adjacency matrix Elements in It represents the influence intensity of the i-th section on the j-th section; Then, the graph convolution method is used to perform cross-section feature fusion, as shown in the following formula (11): （11）； In formula (11), is the adjacency matrix with self-loops added, represents the identity matrix; represent The degree matrix of ; Represents the PDM module Output of layer structure; Represents the PDM module The weight matrix of the layer structure, Represents the PDM module The dimension size of the layer structure input time series feature, Represents the weight matrix The dimension of the output feature after transformation; represents a nonlinear activation function; Represents the PDM module Cross-section fusion results of layer structures; Finally, the PDM module Cross-section fusion results of layer structure With mixed seasonal components , mixed trend component Fusion is performed to obtain the PDM module Output of layer structure , see the following formula (12) for details: （12）； In formula (12), Represents a splicing operation.

8. The multi-objective optimization method for coking and production based on a global optimization strategy according to claim 3, characterized in that: The multi-objective optimization model is solved using a non-dominated sorting genetic algorithm III. That is, the real-time production data of the coking co-production is embedded in a multi-objective optimization function with constraints using the non-dominated sorting genetic algorithm III and solved to obtain a Pareto optimal solution set, any one of which is the optimal solution, thus forming a set of optimal production decision plans.

9. A multi-objective optimization system for coking and production based on a global optimization strategy, characterized by: It includes data acquisition and verification module, global multi-objective optimization engine, collaborative production decision module and multi-section collaborative control module; The data acquisition and verification module is used to collect real-time production data of coking and co-production, normalize the characteristic values ​​of multiple variable parameters in the real-time production data, perform data verification, and upload all verified and retained variable parameters to the global multi-objective optimization engine; The global multi-objective optimization engine includes a trained coking co-production prediction model and a multi-objective optimization model. The coke production and methanol production are predicted based on the trained coking co-production prediction model. The multi-objective optimization model uses a multi-objective optimization function with constraints to perform multi-objective optimization and obtain a Pareto optimal solution set. The collaborative production decision module corresponds the Pareto optimal solution set to the control parameters of all sections of the coking and co-production, and the control parameters of all sections of the coking and co-production form the optimal production decision plan; The multi-section collaborative control module is used to carry out collaborative automated production of multiple sections in coking joint production according to the optimal production decision-making plan, and feed back real-time production data to the data acquisition and verification module.

Images