Coal mine carbon emission accounting management system

Abstract

The invention belongs to the technical field of carbon emission accounting management, and particularly relates to a coal mine carbon emission accounting management system which comprises a data acquisition module, a data processing module, a carbon emission accounting module and a carbon emission management optimization module. The carbon emission accounting module integrates and constructs a high-dimensional variable space, decomposes the high-dimensional variable space into low-dimensional subspaces according to an energy coupling relationship, and constructs a carbon emission calculation model in combination with an improved reinforcement learning algorithm and a PDEs model, so that the dynamic change of carbon emission in the coal mine production process can be comprehensively and accurately reflected, and the carbon emission calculation result is more accurate; the carbon emission management optimization module sets and processes basic constraint conditions, forms a constraint space, reasonably divides and sorts a population by calculating a crowding distance, continuously searches an optimal individual through neighborhood variation, and reasonably distributes high-intensity constraints to different optimization tasks, so that the carbon emission management efficiency of a coal mine system can be effectively improved, and the carbon emission management efficiency of the coal mine system is improved. And the coal mine carbon emission can be reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of carbon emission accounting management, and specifically refers to a coal mine carbon emission accounting management system. Background Art

[0002] The coal mine carbon emission accounting and management system is a technical system for accurate calculation and efficient management of coal mine carbon emissions, which helps the coal mine industry to save energy and reduce emissions. However, the current coal mine carbon emission accounting and management system has the following shortcomings:

[0003] In terms of carbon emission accounting, due to the lack of comprehensive and in-depth process analysis of the coal mine system and the lack of targeted research on the coal mine production links based on energy coupling relationships, key factors are easily overlooked, resulting in inaccurate carbon emission accounting, making it difficult to truly adapt to the dynamic changes in carbon emissions in the coal mine production process, and unable to provide a reliable basis for energy conservation and emission reduction in the coal mine industry;

[0004] In terms of carbon emission management, it is impossible to form a scientific constraint space. When optimizing the carbon emission management plan, it is inefficient in dealing with constraints and searching for the optimal plan. It is impossible to reasonably allocate resources, and it is difficult to adapt to the dynamics and complexity of coal mine production. It is impossible to effectively reduce coal mine carbon emissions, which is not conducive to the sustainable development of the coal mining industry and it is difficult to meet increasingly stringent environmental protection requirements. Summary of the invention

[0005] In response to the above problems, the present invention provides a coal mine carbon emission accounting and management system. The carbon emission accounting module integrates and constructs a high-dimensional variable space, decomposes it into low-dimensional subspaces according to the energy coupling relationship, and constructs a carbon emission calculation model by combining the improved reinforcement learning algorithm and the PDEs model. It can comprehensively and accurately reflect the dynamic changes of carbon emissions in the coal mine production process, and make the carbon emission calculation results more accurate; the carbon emission management optimization module sets and processes basic constraints to form a constraint space, reasonably divides and sorts the population by calculating the crowding distance, continuously searches for the optimal individual through neighborhood variation, and reasonably allocates high-intensity constraints to different optimization tasks, which can effectively improve the carbon emission management efficiency of the coal mine system and help reduce the carbon emissions of coal mines.

[0006] A coal mine carbon emission accounting management system, comprising a data collection module, a data processing module, a carbon emission accounting module and a carbon emission management optimization module;

[0007] The data acquisition module is connected to the data processing module, the data processing module is connected to the carbon emission accounting module, and the carbon emission accounting module is connected to the carbon emission management optimization module.

[0008] The data collection module collects coal mine production activity data and carbon emission coefficients;

[0009] The data processing module performs preprocessing operations on the coal mine production activity data to obtain coal mine data;

[0010] The carbon emission accounting module introduces an improved reinforcement learning algorithm and a PDEs model to construct a carbon emission calculation model, train the carbon emission calculation model, and generate coal mine carbon emission results;

[0011] The carbon emission management optimization module sets basic constraints, combines the basic constraints to form a constraint space, introduces a knowledge transfer strategy, recombines the basic constraints to construct an optimization task, and generates a carbon emission management plan;

[0012] The process of obtaining coal mine data by the data processing module includes the following steps:

[0013] Step D1: Data fusion: Fusion of coal mine production activity data, using cross-validation method to determine the logical relationship between coal mine production activity data of different dimensions;

[0014] Step D2: Data standardization: unify the temporal and spatial scales of coal mine production activity data and convert some coal mine production activity data;

[0015] Step D3: Data optimization: construct a multi-layer autoencoder, train the autoencoder to learn the intrinsic characteristics of the coal mine production activity data, identify the noise and outliers in the coal mine production activity data, and reconstruct the coal mine production activity data;

[0016] Step D4: Data supplement: Generative adversarial network technology is used to build a GAN model to generate simulated data to supplement the coal mine production activity data and obtain coal mine data.

[0017] Further, step D4 comprises the following steps:

[0018] Step D41: construct a GAN model: including a generator and a discriminator, randomly sample noise vectors and input them into the generator, input the coal mine production activity data and the simulated coal mine production activity data into the discriminator at the same time, and the discriminator outputs the probability that the simulated coal mine production activity data is the coal mine production activity data;

[0019] Step D42: Generate simulation data: Use the generator to generate simulation data, use the simulation data to supplement the missing parts of the coal mine production activity data, and comprehensively obtain the coal mine data.

[0020] The carbon emission accounting module generates the coal mine carbon emission results, including the following steps:

[0021] Step A1: Coal mine system modeling: Model the coal mine system, conduct process analysis on the coal mine work links, determine the key factors and abstract variables, use the system dynamics model, integrate the key factors, and construct a high-dimensional variable space;

[0022] Step A2: Decomposition of high-dimensional variable space: Analyze the energy coupling relationship in the coal mine system, extract variable combinations based on the energy coupling relationship, and decompose the high-dimensional variable space into low-dimensional subspaces, each of which corresponds to a coal mine work task;

[0023] Step A3: Carbon emission calculation: Introduce the improved reinforcement learning algorithm combined with the PDEs model, build a carbon emission calculation model, initialize the main network and the super network, use the main network and the super network respectively in each low-dimensional subspace, calculate the carbon emission results, train the carbon emission calculation model, and generate coal mine carbon emission results.

[0024] Furthermore, step A1 includes the following steps:

[0025] Step A11: Variable abstraction: Perform process analysis on each coal mine work link, determine the key factors of each coal mine work link, and abstract the key factors into variables. The coal mine work links include coal mining link, transportation link and lifting link;

[0026] Step A12: Variable integration: Use the system dynamics model to quantitatively describe the relationship between variables, integrate the key factors of coal mine work links, and construct a high-dimensional variable space.

[0027] Further, step A3 includes the following steps:

[0028] Step A31: Introduce an improved reinforcement learning algorithm combined with the PDEs model to construct a carbon emission calculation model. The model includes a main network and a super network. Initialize the main network and the super network. In each low-dimensional subspace, combine the variables and the carbon emission coefficient to construct a carbon emission dynamic change equation.

[0029] The dynamic change equation of carbon emissions is constructed, and the formula used is as follows:

[0030] ;

[0031] ;

[0032] Among them, y represents the state solution, u represents the input function, t and w represent the time variables, Indicates the current state. represents the control input, represents a dynamic function, Represents the objective function, that is, finding the appropriate y and w to minimize the calculation error. represents the parameter vector, represents the instantaneous carbon emission calculation function, and Respectively represent the upper and lower limits of the time interval;

[0033] Step A32: Determine the input information of the hypernetwork, the input information is energy consumption data, the hypernetwork calculates the input information, learns to generate the weights and biases of the main network, the main network processes the variables and carbon emission coefficients, and calculates the carbon emission results;

[0034] Step A33: Set the reward signal, construct the loss function, train the carbon emission calculation model based on the reward signal and the loss function, adjust the weights and bias of the main network, and generate the coal mine carbon emission results.

[0035] The process of generating a carbon emission management plan by the carbon emission management optimization module includes the following steps:

[0036] Step S1: Setting basic constraints: determining the carbon optimization target, setting basic constraints based on the carbon optimization target, and combining the basic constraints to form a constraint space;

[0037] Step S2: Calculate crowding distance: construct decision space and target space, the decision space contains gene information, the target space contains target value, introduce knowledge transfer strategy, combine with multi-task evolutionary optimization algorithm, initialize population and individuals, divide population into frontiers, and calculate crowding distance of individuals on the same frontier;

[0038] Step S3: Perform knowledge transfer: sort the crowding distances, select the top 20% of individuals from each frontier as elite individuals, transfer the gene information of the elite individuals in the decision space and the target value in the target space, obtain the Pareto frontier, and guide the search direction based on the Pareto frontier;

[0039] Step S4: Perform neighborhood mutation: calculate the vector angle between individuals, select individuals to form a neighborhood according to the vector angle, perform mutation operations within the neighborhood, generate the optimal individual, and optimize the search direction based on the optimal individual;

[0040] Step S5: Generate a management plan: select basic constraints and combine them, construct multiple optimization tasks, reasonably allocate high-intensity constraints to different optimization tasks, perform optimization tasks in combination with the search direction, and generate a carbon emission management plan.

[0041] Furthermore, step S1 includes the following steps:

[0042] Step S11: basic constraint condition evaluation: perform strength evaluation on the basic constraint conditions to obtain strength evaluation results, and classify the basic constraint conditions according to the strength evaluation results into high-strength constraints, medium-strength constraints, and low-strength constraints;

[0043] Step S12: Processing basic constraints: using linearization method, relaxation method and penalty function method to process basic constraints and form a constraint space;

[0044] The penalty function method uses the following formula:

[0045] ;

[0046] ;

[0047] in, and is the value of the penalty function, represents the value of the inequality penalty function, represents the value of the equality penalty function, i represents the basic constraint index, represents the penalty factor, x represents the decision variable, represents the basic constraint condition of the i-th inequality, It ensures that the penalty function only penalizes the violation of the constraint. Represents the basic constraints of the ith equation.

[0048] Further, step S2 includes the following steps:

[0049] Step S21: Target dimension division: construct a target space according to the carbon optimization target, and divide the target space into different target dimensions;

[0050] Step S22: frontier division: initialize the population and individuals in the multi-task evolutionary optimization algorithm, divide the population into different frontiers by the non-dominated sorting method, select individuals on the same frontier as the objects for calculating the crowding distance, and individuals on the same frontier do not dominate each other in the target space;

[0051] Step S23: Individual sorting: traverse all target dimensions, sort all individuals on the same frontier according to the target value on the target dimension, and obtain an individual sorting sequence;

[0052] Step S24: Distance calculation: In each target dimension, calculate the individual distance difference in the middle of the individual sorting sequence, sum the individual distance differences of all target dimensions of the same frontier individuals, calculate the crowding distance of each individual, and obtain the crowding distance set of the same frontier, which reflects the distribution crowding degree of each individual in the same frontier in the target space;

[0053] The crowding distance of each individual is calculated using the following formula:

[0054] ;

[0055] Among them, i and j represent individual indexes, represents the crowding distance, k represents the target dimension index, and m represents the number of target dimensions. represents the target value of the i-th individual on the k-th target dimension, represents the target value of the jth individual on the kth target dimension.

[0056] Further, step S4 includes the following steps:

[0057] Step S41: Neighborhood construction: Calculate the vector angle between individuals, select individuals to form neighborhoods based on the vector angle, and construct 10 neighborhoods for each individual in the population;

[0058] Step S42: Global search: randomly select individuals in the neighborhood to perform local mutation operations, guide the search direction, generate new individuals and screen the current best individuals, perform mutation operations on the population, and optimize the search direction based on the current best individuals;

[0059] Individuals are randomly selected in the neighborhood for local mutation operations. The formula used is as follows:

[0060] ;

[0061] in, represents a variant individual, , , represents different individuals in the randomly selected population, and F represents the scaling factor;

[0062] The formula used to perform mutation operation on the population is as follows:

[0063] ;

[0064] in, represents the individual variation at the population level, represents the optimal individual, , Represents different individuals in a randomly selected population.

[0065] The beneficial effects achieved by the present invention are as follows:

[0066] (1) The system uses the carbon emission accounting module to conduct process analysis on each working link of the coal mine system, determine the key factors and abstract them into variables, use the system dynamics model to integrate and construct a high-dimensional variable space, decompose it into low-dimensional subspaces based on the energy coupling relationship, and combine the improved reinforcement learning algorithm with the PDEs model to construct a carbon emission calculation model, which can comprehensively and accurately reflect the dynamic changes of carbon emissions in the coal mine production process, making the carbon emission calculation results more accurate;

[0067] (2) The system uses the carbon emission management optimization module to clarify the carbon optimization target, set and process basic constraints to form a constraint space, generate a carbon emission management plan by calculating the crowding distance, knowledge transfer and neighborhood variation operations, reasonably divide and sort the population when calculating the crowding distance, continuously search for the optimal individual through neighborhood variation, and reasonably allocate high-intensity constraints to different optimization tasks, which effectively improves the carbon emission management efficiency of the coal mine system and helps to reduce the carbon emissions of coal mines. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a module schematic diagram of a coal mine carbon emission accounting and management system proposed by the present invention;

[0069] Figure 2 This is a flow chart of the carbon emission management optimization module proposed in the present invention. DETAILED DESCRIPTION

[0070] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0071] Example 1: Reference Figure 1 ,This embodiment provides a coal mine carbon emission accounting management system, a data acquisition module, a data processing module, a carbon emission accounting module and a carbon emission management optimization module;

[0072] The data acquisition module is connected to the data processing module, the data processing module is connected to the carbon emission accounting module, and the carbon emission accounting module is connected to the carbon emission management optimization module.

[0073] The data collection module collects coal mine production activity data and carbon emission coefficients. The coal mine production activity data includes coal mining volume, coal washing volume, coal transportation volume, carbon emission data, energy consumption data, coal mine equipment information, coal mine geological data and personnel activity data;

[0074] The data processing module performs preprocessing operations on the coal mine production activity data to obtain coal mine data;

[0075] The carbon emission accounting module introduces an improved reinforcement learning algorithm and a PDEs model to construct a carbon emission calculation model, train the carbon emission calculation model, and generate coal mine carbon emission results;

[0076] The carbon emission management optimization module sets basic constraints, combines the basic constraints to form a constraint space, introduces a knowledge transfer strategy, recombines the basic constraints to construct an optimization task, and generates a carbon emission management plan;

[0077] Embodiment 2: This embodiment is based on Embodiment 1, and the process of obtaining coal mine data by the data processing module includes the following steps:

[0078] Step D1: Data fusion: Fusion of coal mine production activity data, using cross-validation method to determine the logical relationship between coal mine production activity data of different dimensions;

[0079] Step D2: Data standardization: unify the temporal and spatial scales of coal mine production activity data and convert some coal mine production activity data;

[0080] In the time dimension, the hourly energy consumption data of different coal mining faces are converted into daily energy consumption data in units of mines. In the spatial dimension, based on the ventilation system and production layout of the coal mine, the carbon transmission and diffusion effects between different coal mine areas are considered, and the carbon emission data of each coal mine area is converted into equivalent carbon emission data based on the entire coal mine area.

[0081] Step D3: Data optimization: construct a multi-layer autoencoder, train the autoencoder to learn the intrinsic characteristics of the coal mine production activity data, identify the noise and outliers in the coal mine production activity data, and reconstruct the coal mine production activity data;

[0082] Step D4: Data supplementation: Generate simulation data to supplement the coal mine production activity data and obtain coal mine data;

[0083] Step D4 comprises the following steps:

[0084] Step D41: construct a GAN model: including a generator and a discriminator, randomly sample noise vectors and input them into the generator, input the coal mine production activity data and the simulated coal mine production activity data into the discriminator at the same time, and the discriminator outputs the probability that the simulated coal mine production activity data is the coal mine production activity data;

[0085] Step D42: Generate simulation data: Use the generator to generate simulation data, use the simulation data to supplement the missing parts of the coal mine production activity data, and comprehensively obtain the coal mine data.

[0086] Embodiment 3: This embodiment is based on Embodiment 2. The process of generating the coal mine carbon emission results by the carbon emission accounting module includes the following steps:

[0087] Step A1: Coal mine system modeling: Model the coal mine system, conduct process analysis on the coal mine work links, determine the key factors and abstract variables, use the system dynamics model, integrate the key factors, and construct a high-dimensional variable space;

[0088] Step A1 includes the following steps:

[0089] Step A11: Variable abstraction: Perform process analysis on each coal mine work link, determine the key factors of each coal mine work link, and abstract the key factors into variables. The coal mine work links include coal mining link, transportation link and lifting link;

[0090] In this embodiment, the coal mining process is first analyzed. The coal mining method is longwall mining. The key factors are the coal cutting speed of the coal mining machine, the moving speed of the hydraulic support, the ventilation volume, the coal seam thickness and the coal seam inclination. The coal cutting speed of the coal mining machine is abstracted as , the moving speed of the hydraulic support is abstracted as , the ventilation volume is abstracted as , the coal seam thickness is abstracted as , the coal seam inclination is abstracted as ;

[0091] Secondly, the transportation link is analyzed. The key factors are the conveying capacity of the conveyor, the belt speed of the conveyor, the length of the transportation line, the slope of the transportation line, the capacity of the mine car, the number of mine cars, the flatness of the track and the radius of the curve. The conveying capacity of the conveyor is abstracted as , the conveyor belt speed is abstracted as , the length of the transport line is abstracted as , the slope of the transportation line is abstracted as , abstracting the minecart capacity into , abstract the number of minecarts into , the track flatness is abstracted as , the curve radius is abstracted as ;

[0092] Finally, the key factors in the lifting process are the lifting capacity of the hoist, the lifting speed of the hoist, the well depth, the container capacity, the strength of the wire rope and the length of the wire rope. The lifting capacity of the hoist is abstracted as , the lifting speed of the elevator is abstracted as , abstracting the well depth into , abstracting the container capacity into , the wire rope strength is abstracted as , the length of the wire rope is abstracted as ;

[0093] Step A12: Variable integration: Use the system dynamics model to quantitatively describe the relationship between variables, integrate the key factors of coal mine work links, and construct a high-dimensional variable space;

[0094] Step A2: Decomposition of high-dimensional variable space: Analyze the energy coupling relationship in the coal mine system. The energy coupling relationship is specifically the correlation between electricity, coal and thermal energy in the process of production, conversion and consumption. According to the energy coupling relationship, extract the variable combination and decompose the high-dimensional variable space into low-dimensional subspaces. Each low-dimensional subspace corresponds to a coal mine work task.

[0095] Step A3: Carbon emission calculation: construct a carbon emission calculation model to generate coal mine carbon emission results;

[0096] Step A3 includes the following steps:

[0097] Step A31: Introduce an improved reinforcement learning algorithm combined with the PDEs model to construct a carbon emission calculation model. The model includes a main network and a super network. Initialize the main network and the super network. In each low-dimensional subspace, combine the variables and the carbon emission coefficient to construct a carbon emission dynamic change equation.

[0098] The dynamic change equation of carbon emissions is constructed, and the formula used is as follows:

[0099] ;

[0100] ;

[0101] Among them, y represents the state solution, u represents the input function, t and w represent the time variables, Indicates the current state. represents the control input, represents a dynamic function, Represents the objective function, that is, finding the appropriate y and w to minimize the calculation error. represents the parameter vector, represents the instantaneous carbon emission calculation function, and Respectively represent the upper and lower limits of the time interval;

[0102] Step A32: Determine the input information of the hypernetwork, the input information is energy consumption data, the hypernetwork calculates the input information, learns to generate the weights and biases of the main network, the main network processes the variables and carbon emission coefficients, and calculates the carbon emission results;

[0103] Step A33: Set the reward signal, construct the loss function, train the carbon emission calculation model based on the reward signal and the loss function, adjust the weights and bias of the main network, and generate the coal mine carbon emission results.

[0104] Embodiment 4: This embodiment is based on Embodiment 2, and the process of generating the coal mine carbon emission results by the carbon emission accounting module includes the following steps:

[0105] Step A1: Coal mine system modeling: Model the coal mine system, conduct process analysis on the coal mine work links, determine the key factors and abstract variables, use the system dynamics model, integrate the key factors, and construct a high-dimensional variable space;

[0106] Step A2: Decomposition of high-dimensional variable space: Analyze the energy coupling relationship in the coal mine system, extract variable combinations based on the energy coupling relationship, and decompose the high-dimensional variable space into low-dimensional subspaces, each of which corresponds to a coal mine work task;

[0107] Step A3: Carbon emission calculation: Combine the PDEs model to build a carbon emission calculation model, calculate the carbon emission results, train the carbon emission calculation model, and generate coal mine carbon emission results.

[0108] Example 5: Reference Figure 2 This embodiment is based on Embodiment 3, wherein the carbon emission management optimization module generates a carbon emission management solution, comprising the following steps:

[0109] Step S1: Setting basic constraints: determining the carbon optimization target, setting basic constraints based on the carbon optimization target, and combining the basic constraints to form a constraint space;

[0110] Step S1 includes the following steps:

[0111] Step S11: basic constraint condition evaluation: perform strength evaluation on the basic constraint conditions to obtain strength evaluation results, and classify the basic constraint conditions according to the strength evaluation results into high-strength constraints, medium-strength constraints, and low-strength constraints;

[0112] Step S12: Processing basic constraints: using linearization method, relaxation method and penalty function method to process basic constraints and form a constraint space;

[0113] The penalty function method uses the following formula:

[0114] ;

[0115] ;

[0116] in, and is the value of the penalty function, represents the value of the inequality penalty function, represents the value of the equality penalty function, i represents the basic constraint index, represents the penalty factor, x represents the decision variable, represents the basic constraint condition of the i-th inequality, It ensures that the penalty function only penalizes the violation of the constraint. represents the basic constraints of the i-th equation;

[0117] Step S2: Calculate crowding distance: construct decision space and target space, introduce knowledge transfer strategy, combine with multi-task evolutionary optimization algorithm, initialize population and individuals, divide population into frontiers, and calculate crowding distance of individuals in the same frontier;

[0118] Step S2 includes the following steps:

[0119] Step S21: Target dimension division: construct a target space according to the carbon optimization target, and divide the target space into different target dimensions;

[0120] Step S22: frontier division: initialize the population and individuals in the multi-task evolutionary optimization algorithm, divide the population into different frontiers by the non-dominated sorting method, select individuals on the same frontier as the objects for calculating the crowding distance, and individuals on the same frontier do not dominate each other in the target space;

[0121] In this embodiment, the frontier contains individual ;

[0122] Step S23: Individual sorting: traverse all target dimensions, sort all individuals on the same frontier according to the target value on the target dimension, and obtain an individual sorting sequence;

[0123] In this embodiment, the target value of target dimension 1 is: individual coal production, sorted as follows: ; The target value of target dimension 2 is: energy consumption, sorted as follows: ;

[0124] Step S24: Distance calculation: In each target dimension, calculate the individual distance difference in the middle of the individual sorting sequence, sum the individual distance differences of all target dimensions of the same frontier individuals, calculate the crowding distance of each individual, and obtain the crowding distance set of the same frontier, which reflects the distribution crowding degree of each individual in the same frontier in the target space;

[0125] The crowding distance of each individual is calculated using the following formula:

[0126] ;

[0127] Among them, i and j represent individual indexes, represents the crowding distance, k represents the target dimension index, and m represents the number of target dimensions. represents the target value of the i-th individual on the k-th target dimension, represents the target value of the jth individual on the kth target dimension;

[0128] Step S3: Perform knowledge transfer: sort the crowding distances, select the top 20% of individuals from each frontier as elite individuals, transfer the gene information of the elite individuals in the decision space and the target value in the target space, obtain the Pareto frontier, and guide the search direction based on the Pareto frontier;

[0129] Step S4: Perform neighborhood mutation: calculate the vector angle between individuals, select individuals to form a neighborhood according to the vector angle, perform mutation operations within the neighborhood, generate the optimal individual, and optimize the search direction based on the optimal individual;

[0130] Step S4 includes the following steps:

[0131] Step S41: Neighborhood construction: Calculate the vector angle between individuals, select individuals to form neighborhoods based on the vector angle, and construct 10 neighborhoods for each individual in the population;

[0132] Step S42: Global search: randomly select individuals in the neighborhood to perform local mutation operations, guide the search direction, generate new individuals and screen the current best individuals, perform mutation operations on the population, and optimize the search direction based on the current best individuals;

[0133] Individuals are randomly selected in the neighborhood for local mutation operations. The formula used is as follows:

[0134] ;

[0135] in, represents a variant individual, , , represents different individuals in the randomly selected population, and F represents the scaling factor;

[0136] The formula used to perform mutation operation on the population is as follows:

[0137] ;

[0138] in, represents the individual variation at the population level, represents the optimal individual, , represents different individuals in a randomly selected population;

[0139] Step S5: Generate a management plan: select basic constraints and combine them, construct multiple optimization tasks, reasonably allocate high-intensity constraints to different optimization tasks, perform optimization tasks in combination with the search direction, and generate a carbon emission management plan.

[0140] Embodiment 6: This embodiment is based on Embodiment 4. The process of generating a carbon emission management plan by the carbon emission management optimization module includes the following steps:

[0141] Step S1: Setting basic constraints: determining the carbon optimization target, setting basic constraints based on the carbon optimization target, and combining the basic constraints to form a constraint space;

[0142] Step S2: Calculate crowding distance: construct decision space and target space, the decision space contains gene information, the target space contains target value, introduce knowledge transfer strategy, combine with multi-task evolutionary optimization algorithm, initialize population and individuals, divide population into frontiers, and calculate crowding distance of individuals on the same frontier;

[0143] Step S3: Perform knowledge transfer: sort the crowding distances, select the top 20% of individuals from each frontier as elite individuals, transfer the gene information of the elite individuals in the decision space and the target value in the target space, obtain the Pareto frontier, and guide the search direction based on the Pareto frontier;

[0144] Step S4: Select basic constraints for combination, construct multiple optimization tasks, reasonably allocate high-intensity constraints to different optimization tasks, perform optimization tasks in combination with the search direction, and generate a carbon emission management plan.

[0145] The present invention and its implementation methods are described above. Such description is not restrictive. If a person skilled in the art is inspired by it and creatively designs structures and implementation methods similar to the technical solution without departing from the purpose of the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A coal mine carbon emission accounting and management system, comprising a data acquisition module and a data processing module, wherein the data acquisition module collects coal mine production activity data and carbon emission coefficients, and the data processing module performs preprocessing operations on the coal mine production activity data to obtain coal mine data; characterized in that: It also includes a carbon emission accounting module and a carbon emission management optimization module; The carbon emission accounting module introduces an improved reinforcement learning algorithm and a PDEs model to construct a carbon emission calculation model, train the carbon emission calculation model, and generate coal mine carbon emission results; The carbon emission management optimization module sets basic constraints, combines the basic constraints to form a constraint space, introduces a knowledge transfer strategy, recombines the basic constraints to construct an optimization task, and generates a carbon emission management plan; The carbon emission accounting module generates the coal mine carbon emission results, including the following steps: Step A1: Coal mine system modeling: Model the coal mine system, conduct process analysis on the coal mine work links, determine key factors and abstract variables, and use the system dynamics model to construct a high-dimensional variable space; Step A2: Decomposition of high-dimensional variable space: Analyze the energy coupling relationship in the coal mine system, extract variable combinations based on the energy coupling relationship, and decompose the high-dimensional variable space into low-dimensional subspaces; Step A3: Carbon emission calculation: Introduce the improved reinforcement learning algorithm combined with the PDEs model, build a carbon emission calculation model, initialize the main network and the super network, use the main network and the super network respectively in each low-dimensional subspace, calculate the carbon emission results, train the carbon emission calculation model, and generate coal mine carbon emission results.

2. A coal mine carbon emission accounting and management system according to claim 1, characterized in that: Step A3 includes the following steps: Step A31: Introduce an improved reinforcement learning algorithm combined with the PDEs model to construct a carbon emission calculation model. The model includes a main network and a super network. Initialize the main network and the super network. In each low-dimensional subspace, combine the variables and the carbon emission coefficient to construct a carbon emission dynamic change equation. Step A32: Determine the input information of the hypernetwork, the input information is energy consumption data, the hypernetwork calculates the input information, learns to generate the weights and biases of the main network, the main network processes the variables and carbon emission coefficients, and calculates the carbon emission results; Step A33: Set the reward signal, construct the loss function, train the carbon emission calculation model based on the reward signal and the loss function, adjust the weights and bias of the main network, and generate the coal mine carbon emission results.

3. A coal mine carbon emission accounting and management system according to claim 1, characterized in that: The process of generating a carbon emission management plan by the carbon emission management optimization module includes the following steps: Step S1: Setting basic constraints: determining the carbon optimization target, setting basic constraints based on the carbon optimization target, and combining the basic constraints to form a constraint space; Step S2: Calculate crowding distance: construct decision space and target space, the decision space contains gene information, the target space contains target value, introduce knowledge transfer strategy, combine with multi-task evolutionary optimization algorithm, initialize population and individuals, divide population into frontiers, and calculate crowding distance of individuals on the same frontier; Step S3: Perform knowledge transfer: sort the crowding distances, select some individuals from each frontier as elite individuals, transfer the gene information of the elite individuals in the decision space and the target value in the target space, obtain the Pareto frontier, and guide the search direction based on the Pareto frontier; Step S4: Perform neighborhood mutation: calculate the vector angle between individuals, select individuals to form a neighborhood according to the vector angle, perform mutation operations within the neighborhood, generate the optimal individual, and optimize the search direction based on the optimal individual; Step S5: Generate a management plan: select basic constraints and combine them, construct multiple optimization tasks, reasonably allocate high-intensity constraints to different optimization tasks, perform optimization tasks in combination with the search direction, and generate a carbon emission management plan.

4. A coal mine carbon emission accounting and management system according to claim 3, characterized in that: Step S2 includes the following steps: Step S21: Target dimension division: construct a target space according to the carbon optimization target, and divide the target space into different target dimensions; Step S22: frontier division: initialize the population and individuals, divide the population into different frontiers, select individuals on the same frontier as objects for calculating crowding distance, and individuals on the same frontier do not dominate each other in the target space; Step S23: Individual sorting: traverse all target dimensions, sort all individuals on the same frontier according to the target value on the target dimension, and obtain an individual sorting sequence; Step S24: Distance calculation: On each target dimension, calculate the individual distance difference in the middle of the individual sorting sequence, sum the individual distance differences of all target dimensions of the same frontier individuals, calculate the crowding distance of each individual, and obtain the crowding distance set of the same frontier.

5. A coal mine carbon emission accounting and management system according to claim 3, characterized in that: Step S4 includes the following steps: Step S41: neighborhood construction: calculating the vector angle between individuals, and selecting individuals to form a neighborhood according to the vector angle; Step S42: Global search: Randomly select individuals in the neighborhood to perform local mutation operations, guide the search direction, generate new individuals and screen the current best individual, perform mutation operations on the population, and optimize the search direction based on the current best individual.

Images