A method and system for industrial production operation optimization based on graph theory

Abstract

The present invention relates to the technical field of industrial production optimization, specifically a graph-theory-based industrial production operation optimization method and system, comprising: 1. collecting operating parameters, fly ash characteristics, energy and power data, and unit price data of different energy types in industrial production processes of different industries; 2. obtaining carbon emissions, combustion efficiency, and energy costs corresponding to each operating condition and process energy consumption; 3. establishing a relationship map of independent variables and dependent variables based on graph theory; 4. solving the optimal carbon emissions, energy costs, and efficiency corresponding to a selected set of independent variables in the map based on the minimum path principle; 5. finding the operating parameter distribution corresponding to the optimal carbon emissions, energy costs, and efficiency and satisfying stable operation constraints, and then optimizing industrial production operations using the operating parameter distribution. The present invention can significantly save experimental costs, improve research efficiency, and provide a scientific basis for energy conservation, carbon reduction, cost reduction, and efficiency improvement in the industrial field.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial production optimization, and in particular to an industrial production operation optimization method and system based on graph theory. Background Art

[0002] For high-energy-consuming production enterprises like power plants and cement plants, optimizing operating parameters to minimize energy costs, carbon emissions, and maximize fuel combustion efficiency is crucial for production and operations. This not only significantly improves operational economics but also serves as a fundamental path to a green, low-carbon transition. Typically, during operational scheduling, enterprises rely on the experience of operators, which involves trial and error, time, and cost, and lacks precise and professional guidance.

[0003] Therefore, there is an urgent need for an industrial production operation optimization method to address problems such as long trial and error time and high trial and error costs. Summary of the Invention

[0004] The present invention provides an industrial production operation optimization method and system based on graph theory to solve the technical problems mentioned in the background technology.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] The present invention provides an industrial production operation optimization method based on graph theory, comprising the following steps:

[0007] S1. Collect operating parameters, fly ash characteristics, energy and power data, and unit price data of different energy types in industrial production processes of different industries;

[0008] S2. Obtain the carbon emissions corresponding to each operating condition using the carbon monitoring method or carbon accounting method; calculate the combustion efficiency corresponding to each operating condition based on the industrial analysis results of fly ash characteristics; and calculate the energy cost corresponding to each operating condition based on the unit price data of different energy types and energy consumption;

[0009] S3. Using various operating parameters as independent variables and carbon emissions, energy costs, and combustion efficiency as dependent variables, establish a relationship graph between the independent and dependent variables based on graph theory.

[0010] S4. Using simulated annealing algorithm and ant colony algorithm and based on the minimum path principle, solve the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to a selected set of independent variables in the graph;

[0011] S5. Find the operating condition parameter distribution that corresponds to the optimal carbon emissions, energy cost, and combustion efficiency and satisfies the stable operation constraints, and then use the solved operating condition parameters to optimize industrial production operations.

[0012] Furthermore, the operating parameters in the industrial production process include one or more of the following: air distribution ratio, raw fuel feeding rate, fuel particle size, fuel moisture content, temperature and pressure; wherein the air distribution ratio includes primary air ratio, secondary air ratio and burnout air ratio;

[0013] The energy and power data include one or more of the following: energy consumption type by process, energy consumption by energy type by process, total energy consumption, and total electricity consumption;

[0014] The energy cost includes cost data required for energy consumption by type.

[0015] Furthermore, the carbon monitoring method or carbon accounting method in S2 to obtain the carbon emissions corresponding to each operating condition specifically includes the following steps:

[0016] S21: Determine whether the enterprise has installed carbon monitoring equipment. If yes, proceed to S22; otherwise, proceed to S23.

[0017] S22. Use carbon monitoring methods to obtain carbon emissions corresponding to energy consumption in each operating condition, i.e., use the company's carbon monitoring equipment to obtain carbon emissions corresponding to energy consumption in each operating condition and process;

[0018] S23. Use the carbon accounting algorithm to obtain the carbon emissions corresponding to the energy consumption of each operating condition based on energy and power data, that is, use the carbon emission factor method to calculate the carbon emissions corresponding to the energy consumption of each operating condition and process.

[0019] Furthermore, the S3 specifically includes the following steps:

[0020] S31. The independent variables including various operating parameters and the dependent variables including carbon emissions, energy costs, and combustion efficiency are respectively used as vertices of the relationship graph;

[0021] S32. Based on the relationship between each vertex, use lines to connect the corresponding vertices to obtain the edges of the relationship graph; at this point, a relationship graph between the independent variable and the dependent variable is obtained.

[0022] Furthermore, the S4 specifically includes the following steps:

[0023] S41. First, a set of independent variables is given, the set of independent variables including selected values ​​of various operating condition parameters; then, each independent variable in the set of independent variables is divided into multiple parts according to the equal-width binning method, and the value range of each independent variable is determined;

[0024] S42. Select the qth value of the i-th independent variable from the set of independent variables in S41. , and then select the value from the range of the value of the i-th independent variable Nearest neighbor value ; Then calculate the value and adjacent values The weight distance of

[0025] ;

[0026] in, Indicates the qth value of the i-th independent variable and adjacent independent variable values The weight distance of

[0027] S43. Without considering other independent variables in a set of independent variables in S41, obtain the value of the independent variable according to the relationship diagram Match the frequency of each dependent variable and take the value of the independent variable The frequency of each dependent variable is named weight ; Then according to the weight and the weight of the corresponding dependent variable itself Find the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of

[0028] ;

[0029] in, Indicates the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of represents the kth value of the jth dependent variable; is the value of the i-th independent variable Match the value of the jth dependent variable The probability matching weight of , j represents the number of dependent variables, and k is the k-th value of dependent variable j;

[0030] S44, looping S42 and S43 until the weighted distance between the value of each independent variable in the set of independent variables given in S41 and the value of the adjacent independent variable and the weighted distance between the value of each independent variable in the set of independent variables given in S41 and the value of the corresponding dependent variable are obtained;

[0031] Then, the weighted distances between the value of each independent variable and the value of the adjacent independent variable in the set of independent variables given in S41 are summed to obtain the sum of the weighted distances. , the calculation formula is as follows:

[0032] ;

[0033] Where p is the number of all independent variables;

[0034] Sum the weighted distances between the value of each independent variable and the value of the corresponding dependent variable in a set of independent variables given in S41 to obtain the sum of the weighted distances , the calculation formula is as follows:

[0035] ;

[0036] S45, then based on the sum of weighted distances Sum of weighted distances Construct the objective function;

[0037] S46. Use simulated annealing algorithm and ant colony algorithm to solve the objective function, optimize the dependent variables, and then obtain the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to the set of independent variables selected in S41.

[0038] Furthermore, the objective function in S45 is specifically as follows:

[0039] .

[0040] Furthermore, the S46 specifically includes the following steps:

[0041] S461. Using a set number of ants to search for an optimal set of dependent variables corresponding to the set of independent variables selected in S41 in a relationship map between independent variables and dependent variables; each set of dependent variables includes selected values ​​of carbon emissions, energy costs, and combustion efficiency;

[0042] S462, calculate the probability of ant s solving the next set of dependent variable combinations V;

[0043] S463. In each iteration, the ant can only modify the value of one unit of the dependent variable in a set of dependent variables in S41. Based on this, the probability of the next set of dependent variable combinations V is converted into a cumulative probability distribution, and the cumulative probability of the dependent variable value being increased or decreased by one unit is obtained. ; The unit is the smallest division unit used in the S41 division process; the dependent variable combination includes three dependent variables, namely carbon emissions, energy costs, and combustion efficiency;

[0044] S464, according to the roulette algorithm, the roulette wheel is divided into 6 parts, and a random number r is randomly generated. , according to the random number r and the cumulative probability , find satisfaction When the random number lands on the first portion of the roulette wheel, it means the first dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. When the random number lands on the second portion, it means the first dependent variable is adjusted down by one unit, while the other two dependent variables remain unchanged. When the random number lands on the third portion, it means the second dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. And so on. The roulette wheel algorithm is used to determine the next optimization direction of each dependent variable.

[0045] S465, looping through S462 to S464 until a preset iteration stop condition is reached. During each iteration, the pheromones on all paths are volatilized according to a set ratio to achieve pheromone update. After the update, the path with the higher pheromone content is more likely to be selected by the ant;

[0046] In addition, in each round of iteration, if the new path is better, that is, the difference between the current solution and the historical optimal solution , then the ants will directly search for the optimal solution along the new path, otherwise, they will set the probability Decide whether to solve the optimization along this path;

[0047] When the number of iterations reaches the set number, the iteration ends, and the optimal carbon emissions, energy costs, and combustion efficiency corresponding to the set of independent variables selected in S41 are obtained.

[0048] Furthermore, the calculation formula for the probability of the next group of ants s solving the dependent variable combination V in S462 is:

[0049] ;

[0050] in, and are the parameters set, which are the importance of pheromones and the weight of heuristic information; u, v, and l are the three node positions of the ant, which correspond to the dependent variable combination U, dependent variable combination V, and dependent variable combination L respectively. U is the value combination of this round of iteration, V is the value combination selected for the next round of iteration, and L is any combination in the candidate action set of the ant's next step; represents the dependent variable value combination of ant s at node u, It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination V; It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination L; Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations L, Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations V, The calculation formula is:

[0051] ;

[0052] Where, Indicates the weighted distance of the dependent variable taking the dependent variable combination U; Indicates the weighted distance of the dependent variable from the dependent variable combination V;

[0053] The cumulative probability in S463 The calculation formula is:

[0054] ;

[0055] in, represents action a, i.e. the value of the dependent variable is adjusted up / down by one unit; It represents the corresponding probability of the ant taking any combination of dependent variables in the candidate combinations.

[0056] Furthermore, the pheromone volatilization in S465 is manifested in the following forms:

[0057] ;

[0058] in, Indicates the The pheromone of the second iteration; Indicates the The pheromone of the second iteration; represents the pheromone volatilization rate; h is the total number of ants; It means the pheromones released by ants according to their own constructed interpretation;

[0059] The calculation formula is as follows:

[0060] ;

[0061] Where Q is a constant, Indicates the difference between the current solution and the historical optimal solution, , is the path length of ant s; is the historical optimal solution; T is the unit temperature, which decreases with the number of iterations; It is the length of a series of connected sequences formed in a complete iteration process, starting from the starting point and selecting edges or nodes step by step according to the algorithm's path selection rules;

[0062] The calculation formula of T is as follows:

[0063] ;

[0064] in, Indicates the The unit temperature of the iteration; For the The unit temperature of the iteration; is the temperature reduction coefficient.

[0065] Another aspect of the present invention provides an industrial production operation optimization system based on graph theory, which includes a computer, wherein the computer device is programmed or configured to execute the industrial production operation optimization method.

[0066] Beneficial effects of the present invention:

[0067] The present invention discloses an industrial production operation optimization method based on graph theory. Compared with traditional industrial production operation optimization methods, the present invention can greatly save experimental costs, improve research efficiency, give full play to the role of energy digitization, and provide theoretical support for increasing production, energy saving, carbon reduction, cost reduction and efficiency improvement in high-energy-consuming or industrial industries such as the power industry, cement industry, and steel industry. It can provide a scientific basis for energy conservation, carbon reduction, cost reduction and efficiency improvement in the industrial field. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is a flowchart of the industrial production operation optimization method of the present invention;

[0069] Figure 2 Graph showing the relationship between the independent variables and the dependent variables in the embodiment of the present invention. DETAILED DESCRIPTION

[0070] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The accompanying drawings illustrate preferred embodiments of the present invention. However, the present invention may be implemented in many other forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present disclosure.

[0071] Reference Figure 1 , the embodiment of the present application provides an industrial production operation optimization method based on graph theory, comprising the following steps:

[0072] S1. Collect operating parameters, fly ash characteristics, energy and power data, and unit price data of different energy types in industrial production processes of different industries;

[0073] S2. Obtain the carbon emissions corresponding to each operating condition using the carbon monitoring method or carbon accounting method (calculated based on energy and power data); calculate the combustion efficiency corresponding to each operating condition based on the industrial analysis results of fly ash characteristics; and calculate the energy cost corresponding to each operating condition based on the unit price data of different energy types and energy consumption.

[0074] S3. Using various operating parameters as independent variables and carbon emissions, energy costs, and combustion efficiency as dependent variables, establish a relationship graph between the independent and dependent variables based on graph theory.

[0075] S4. Using simulated annealing algorithm and ant colony algorithm and based on the minimum path principle, solve the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to a selected set of independent variables in the graph;

[0076] S5. Find the operating parameter distribution that corresponds to the optimal carbon emissions, energy cost, and combustion efficiency while satisfying the stable operation constraints. Then, use the solved operating parameters to optimize industrial production operations. The stable operation constraint is that the temperature variation at each point remains within ±10%.

[0077] Compared with traditional industrial production operation optimization methods, this invention can greatly save experimental costs, improve research efficiency, give full play to the role of energy digitalization, and provide theoretical support for increasing production, energy saving, carbon reduction, cost reduction and efficiency improvement in high-energy-consuming or industrial industries such as the power industry, cement industry, and steel industry. It can provide a scientific basis for energy conservation, carbon reduction, cost reduction and efficiency improvement in the industrial field.

[0078] In some embodiments, the operating condition parameters in the industrial production process include one or more of air distribution ratio, raw fuel feeding rate, fuel particle size, fuel moisture content, temperature and pressure; wherein the air distribution ratio includes primary air ratio, secondary air ratio and burnout air ratio;

[0079] The energy and power data include one or more of the following: energy consumption type by process, energy consumption by energy type by process, total energy consumption, and total electricity consumption;

[0080] The energy cost includes cost data required for energy consumption by type.

[0081] In some embodiments, the carbon monitoring method or carbon accounting method in S2 to obtain the carbon emissions corresponding to each operating condition specifically includes the following steps:

[0082] S21: Determine whether the enterprise has installed carbon monitoring equipment. If yes, proceed to S22; otherwise, proceed to S23.

[0083] S22. Use carbon monitoring methods to obtain the carbon emissions corresponding to the energy consumption of each operating condition, that is, use the company's carbon monitoring equipment to obtain the carbon emissions corresponding to the energy consumption of each operating condition;

[0084] S23. Use the carbon accounting method to obtain the carbon emissions corresponding to the energy consumption of each working condition, that is, use the carbon emission factor method to calculate the carbon emissions corresponding to the energy consumption of each working condition.

[0085] In some embodiments, S3 specifically includes the following steps:

[0086] S31. The independent variables including various operating parameters and the dependent variables including carbon emissions, energy costs, and combustion efficiency are respectively used as vertices of the relationship graph;

[0087] S32. Based on the relationship between each vertex, use lines to connect the corresponding vertices to obtain the edges of the relationship graph; at this point, a relationship graph between the independent variable and the dependent variable is obtained.

[0088] In some embodiments, the S4 specifically includes the following steps:

[0089] S41. First, a set of independent variables is given, where the set of independent variables includes selected values ​​of various operating condition parameters;

[0090] Then, each independent variable in the group of independent variables is divided into multiple parts according to the equal-width binning method, and the value range of each independent variable is determined;

[0091] Specifically, if the length of the independent variable's value interval is A, and the independent variable is divided into z parts according to the idea of ​​equal-width binning, then the independent variable's value range is [m1,m2,m3,...mz]. After equal-width binning, each independent variable can match the corresponding dependent variable in the historical data;

[0092] S42. Select the qth value of the i-th independent variable from the set of independent variables in S41. , and then select the value from the range of the value of the i-th independent variable Nearest neighbor value ; Then calculate the value and adjacent values The weight distance of

[0093] ;

[0094] in, Indicates the qth value of the i-th independent variable and adjacent values The weight distance of

[0095] S43, without considering other independent variables in a set of independent variables in S41, obtain the value according to the relationship diagram Match the frequency of each dependent variable and take the value The frequency of each dependent variable is named weight ; Then according to the weight and the weight of the corresponding dependent variable itself Find the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of

[0096] ;

[0097] in, Indicates the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of represents the kth value of the jth dependent variable; is the value of the i-th independent variable Match the value of the jth dependent variable The probability matching weight of , j represents the number of dependent variables, and k is the k-th value of dependent variable j;

[0098] S44, looping S42 and S43 until the weighted distance between the value of each independent variable and the adjacent value in the set of independent variables given in S41 and the weighted distance between the value of each independent variable and the value of the corresponding dependent variable in the set of independent variables given in S41 are obtained;

[0099] Then, the weighted distances between the value of each independent variable and its adjacent values ​​in the set of independent variables given in S41 are summed to obtain the sum of the weighted distances. , the calculation formula is as follows:

[0100] ;

[0101] Where p is the number of all independent variables;

[0102] Sum the weighted distances between the value of each independent variable and the value of the corresponding dependent variable in a set of independent variables given in S41 to obtain the sum of the weighted distances , the calculation formula is as follows:

[0103] ;

[0104] S45, then based on the sum of weighted distances Sum of weighted distances Construct the objective function;

[0105] S46. Use simulated annealing algorithm and ant colony algorithm to solve the objective function, optimize the dependent variables, and then obtain the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to the set of independent variables selected in S41.

[0106] Specifically, the ant colony algorithm is responsible for exploring the large-scale solution space, and the simulated annealing algorithm performs local fine-tuning on the candidate solutions; the temperature parameter of the simulated annealing algorithm is used to regulate the pheromone volatilization rate of the ant colony algorithm to enhance the applicability of the algorithm.

[0107] In some embodiments, the objective function in S45 is specifically as follows:

[0108] .

[0109] In some embodiments, the S46 specifically includes the following steps:

[0110] S461. Using a set number of ants to search for an optimal set of dependent variables corresponding to the set of independent variables selected in S41 in a relationship map between independent variables and dependent variables; each set of dependent variables includes selected values ​​of carbon emissions, energy costs, and combustion efficiency;

[0111] In the process of solving min(F(x)), first select the independent variable , using the greedy method to select independent variables The closest set of dependent variables , As the initial solution, its distance is ;

[0112] S462, calculate the probability of ant s solving the next set of dependent variable combinations V;

[0113] S463. In each iteration, the ant can only modify the value of one unit of the dependent variable in a set of dependent variables in S41. Based on this, the probability of the next set of dependent variable combinations V is converted into a cumulative probability distribution, and the cumulative probability of the dependent variable value being increased or decreased by one unit is obtained. ; The unit is the smallest division unit used in the S41 division process; the dependent variable combination includes three dependent variables, namely carbon emissions, energy costs, and combustion efficiency;

[0114] S464, according to the roulette algorithm, the roulette wheel is divided into 6 parts, and a random number r is randomly generated. , according to the random number r and the cumulative probability , find satisfaction When the random number lands on the first portion of the roulette wheel, it means the first dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. When the random number lands on the second portion, it means the first dependent variable is adjusted down by one unit, while the other two dependent variables remain unchanged. When the random number lands on the third portion, it means the second dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. And so on. The roulette wheel algorithm is used to determine the next optimization direction of each dependent variable.

[0115] S465, looping through S462 to S464 until a preset iteration stop condition is reached. During each iteration, the pheromones on all paths are volatilized according to a set ratio to achieve pheromone update. After the update, the path with the higher pheromone content is more likely to be selected by the ant;

[0116] In addition, in each round of iteration, if the new path is better, that is, the difference between the current solution and the historical optimal solution , then the ants will directly search for the optimal solution along the new path, otherwise, they will set the probability Decide whether to solve the optimization along this path;

[0117] When the number of iterations reaches the set number, the iteration ends, and the optimal carbon emissions, energy costs, and combustion efficiency corresponding to the set of independent variables selected in S41 are obtained.

[0118] In some embodiments, the calculation formula for the probability of the next set of dependent variable combinations V to be solved by ant s in S462 is:

[0119] ;

[0120] in, and are the parameters set, which are the importance of pheromones and the weight of heuristic information; u, v, and l are the three node positions of the ant, which correspond to the dependent variable combination U, dependent variable combination V, and dependent variable combination L respectively. U is the value combination of this round of iteration, V is the value combination selected for the next round of iteration, and L is any combination in the candidate action set of the ant's next step; represents the dependent variable value combination of ant s at node u, It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination V; It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination L; Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations L, Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations V, The calculation formula is:

[0121] ;

[0122] Where, Indicates the weighted distance of the dependent variable taking the dependent variable combination U; Indicates the weighted distance of the dependent variable from the dependent variable combination V;

[0123] The cumulative probability in S463 The calculation formula is:

[0124] ;

[0125] in, represents action a, i.e. the value of the dependent variable is adjusted up / down by one unit; It represents the corresponding probability of the ant taking any combination of dependent variables in the candidate combinations.

[0126] In some embodiments, the pheromone volatilization in S465 is manifested in the form of:

[0127] ;

[0128] in, Indicates the The pheromone of the second iteration; Indicates the The pheromone of the second iteration; represents the pheromone volatilization rate; is the total number of ants; It means the pheromones released by ants according to their own constructed interpretation;

[0129] The calculation formula is as follows:

[0130] ;

[0131] Where Q is a constant, Indicates the difference between the current solution and the historical optimal solution, , is the path length of ant s; is the historical optimal solution; T is the unit temperature, which decreases with the number of iterations; It is the length of a series of connections (edges) formed in a complete iteration process, starting from the starting point, selecting edges or nodes step by step according to the algorithm's path selection rules;

[0132] The calculation formula of T is as follows:

[0133] ;

[0134] in, Indicates the The unit temperature of the iteration; For the The unit temperature of the iteration; is the temperature reduction coefficient, usually taken as 0.999.

[0135] In some embodiments, the step S5 specifically includes the following steps:

[0136] Find the operating parameters and energy consumption distribution corresponding to the specified carbon emissions y1, energy cost y2 and efficiency y3 and satisfying the stable operation constraints;

[0137] Specifically, the independent variables in S4 are converted into dependent variables, and the dependent variables are converted into independent variables. By using the same method as S4, the operating condition parameters and energy consumption distribution corresponding to the specified carbon emissions y1, energy cost y2 and efficiency y3 and satisfying the stable operation constraints can be solved.

[0138] For easier understanding, the following specific examples are used for illustration:

[0139] Assuming the power plant's operating parameters and ranges include primary air ratio (30%-40%), secondary air ratio (30%-40%), overburn air ratio (20%-40%), furnace temperature (850°C-1100°C), furnace pressure (20kPa-30kPa), and coal pulverized particle size (20-50μm), the importance of combustion efficiency, carbon emissions, and fuel costs is weighted at 25%, 40%, and 35%, respectively. It is assumed that power plant combustion efficiency typically ranges from 90% to 99%. While maintaining the same power generation output, carbon emissions are assumed to range from 5,000 tons to 6,000 tons, and energy costs are assumed to range from 10,000 to 15,000 yuan.

[0140] First, divide the power plant's independent variables into equal parts: primary air, secondary air, and overburned air are each divided into 100 equal parts, furnace temperature is divided into 2500 equal parts, furnace pressure is divided into 100 equal parts, and pulverized coal particle size is divided into 300 equal parts. Combustion efficiency, carbon emissions, and energy costs are divided into 90, 1000, and 5000 equal parts, respectively. Each pulverized coal combustion efficiency value is 0.1%, each carbon emission value is 1 ton, and each energy cost value is 1 yuan.

[0141] Given a set of independent variables: the primary air ratio is 35.3%, the secondary air ratio is 30.5%, the burnout air ratio is 34.5%, the furnace temperature is 950.5℃, the furnace pressure is 25kPa, and the coal powder particle size is 45μm, find the dependent variable combination that best matches this set of independent variables and is closest to reality.

[0142] The first step of the two-step weighting method: When the primary air ratio is 35.3%, the weighted distance can be calculated with the adjacent equal-values ​​of 35%, 36%, or finer granularity. When the secondary air ratio is 30.5%, the weighted distance can be calculated with the adjacent equal-values ​​of 30%, 31%, or finer granularity. When the overburning air ratio is 34.5%, the weighted distance can be calculated with the adjacent equal-values ​​of 34%, 35%, or finer granularity. When the furnace temperature is 950.5℃, the weighted distance can be calculated with the adjacent equal-values ​​of 949℃, 950℃, or other adjacent values. The weighted distances of other variables are calculated in the same way.

[0143] Step 2: Assuming that other independent variables are not considered, according to the historical relationship map, the frequency of each independent variable matching each dependent variable can be obtained. For example, when the primary air ratio is 35.3%, the combustion efficiency corresponding to the frequency of 30% is 95%, the combustion efficiency corresponding to the frequency of 20% is 93%, the combustion efficiency corresponding to the frequency of 35% is 98%, and the combustion efficiency corresponding to the frequency of 15% is 97%. Similarly, without considering other variables, when the secondary air ratio is 30.5%, the combustion efficiency corresponding to the frequency of 20% is 95%, the combustion efficiency corresponding to the frequency of 10% is 93%, the combustion efficiency corresponding to the frequency of 50% is 98%, and the combustion efficiency corresponding to the frequency of 20% is 97%. Similarly, the weights of other independent variables matching with dependent variables are obtained. The weight distance between the independent variable and the dependent variable is equal to the weight and the weight of each dependent variable itself Repeat this process for each independent variable.

[0144] The third step: Taking the minimization of the product of the two-step weights as the objective function, the simulated annealing algorithm and the ant colony algorithm are used to optimize the dependent variables. The ant colony algorithm is responsible for exploring the large-scale solution space, and the simulated annealing algorithm performs local fine-tuning on the candidate solutions. The temperature parameter of the simulated annealing algorithm is used to control the pheromone volatilization rate of the ant colony algorithm to enhance the applicability of the algorithm.

[0145] ;

[0146] ;

[0147] in, yes match The probability matching weight comes from the statistical results of the original data set, p is the number of all independent variables, and j is the number of dependent variables.

[0148] Assume that there are 10,000 sets of original data in the historical data. The original data contains all independent variables and all dependent variables. In the process of solving min(F(x)), first select the independent variable , using the greedy method to select independent variables The closest set of dependent variables As the initial solution, its distance is The solution optimization process can then be divided into two steps: path construction and pheromone update.

[0149] Path construction: For each ant s, the path memory vector R(s) records the weighted distances between all independent variables and dependent variables in the order of access. and , let the current set of dependent variable values ​​solved by ant s be U, and the next set of dependent variable values ​​be V. represents the dependent variable value combination of ant s at node u, represents the pheromone concentration from the dependent variable value U to the dependent variable value V, It is a heuristic function, usually taking the inverse of the distance. In the invention book, the function value is Let the probability of ant s taking the dependent variable V be

[0150] ;

[0151] in, and are the parameters set, which are the importance of pheromone and the weight of heuristic information. u, v, l are the node locations of ants.

[0152] In this invention, each ant can only modify the value of one unit of the dependent variable in the dependent variable set, that is, it can only select one dependent variable to add or subtract one unit. Therefore, since each dependent variable cannot exceed its value range, the dependent variable value combination (or candidate set) The maximum number of is 6, let it be n, . Convert the probability into cumulative probability distribution, the cumulative probability is

[0153] ;

[0154] in is the cumulative probability of the a-th action, when a is n, .

[0155] According to the roulette algorithm, a random number r is first generated. , according to the random number r and the cumulative probability , find satisfaction , that is, select a as the next optimization direction of the dependent variable solution I.

[0156] Pheromone Update:

[0157] At initialization, let all pheromones be , change the value to a constant value. In each round of iteration, the pheromone on all paths will evaporate. Let its volatility be At the same time, the ants release pheromones according to their own interpretation, making it , For the Ants, that is

[0158] ;

[0159] ;

[0160] ;

[0161] Where, ρ is the pheromone volatilization rate; Determined by the performance of ant s in solving U→V; Q is a constant, is the path length of ant s; (the difference between the current solution and the historical optimal solution); is the cooling coefficient, T is the unit temperature, which decreases with the number of iterations; h is the total number of ants.

[0162] After the ant chooses the path, the probability Decide whether to keep the path. If the new path is better ( ), accept directly, if worse ( ), with probability accept.

[0163] When the number of iterations reaches n, the historical optimal solution is output , which is the optimal dependent variable combination corresponding to the given independent variables, can be obtained when the cooling rate is 0.97, the pheromone weight is 1, the heuristic weight is 2.0, 50 ants are selected in each generation, and after 200 generations of iteration, it has basically converged. The improvement of subsequent iterations is much less than 0.1%. The optimal solution obtained is (97.15%, 5512.83 tons, 12376.92 yuan), and the total minimum weight distance is 0.2874.

[0164] Similarly, given a certain operating condition, such as a combustion efficiency of 98%, CO2 emissions of 6,000 tons, and an energy cost of 13,000 yuan, the optimal independent variable combination (37.3%, 33.2%, 38.9%, 1025°C, 27.8 kPa, 47 μm) is obtained, with a total minimum weighted distance of approximately 0.2036. The more historical data, the more accurate the optimization results. Similarly, under given operating conditions, the optimal operating parameters can be obtained for reducing emissions and energy costs by a certain percentage while maintaining combustion efficiency. This is done by solving the problem given some of the dependent variable values.

[0165] Another aspect of the present invention provides an industrial production operation optimization system based on graph theory, which includes a computer, wherein the computer device is programmed or configured to execute the industrial production operation optimization method.

[0166] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for optimizing industrial production operations based on graph theory, characterized in that: The steps include: S1. Collect operating parameters, fly ash characteristics, energy and power data, and unit price data of different energy types in industrial production processes of different industries; S2. Obtain the carbon emissions corresponding to each operating condition using a carbon monitoring method or a carbon accounting method; and calculate the combustion efficiency corresponding to each operating condition based on the industrial analysis results of fly ash characteristics; Calculate the energy cost corresponding to each working condition based on the unit price data of different energy types and energy consumption; S3. Using various operating parameters as independent variables and carbon emissions, energy costs, and combustion efficiency as dependent variables, establish a relationship graph between the independent and dependent variables based on graph theory. S4. Using simulated annealing algorithm and ant colony algorithm and based on the minimum path principle, solve the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to a selected set of independent variables in the graph; S5. Find the operating condition parameter distribution that corresponds to the optimal carbon emissions, energy cost, and combustion efficiency and satisfies the stable operation constraints, and then use the solved operating condition parameters to optimize industrial production operations.

2. The industrial production operation optimization method based on graph theory according to claim 1 is characterized in that: The operating parameters in the industrial production process include one or more of the following: air distribution ratio, raw fuel feeding rate, fuel particle size, fuel moisture content, temperature and pressure; wherein the air distribution ratio includes primary air ratio, secondary air ratio and burnout air ratio; The energy and power data include one or more of the following: energy consumption type by process, energy consumption by energy type by process, total energy consumption, and total electricity consumption; The energy cost includes cost data required for energy consumption by type.

3. The industrial production operation optimization method based on graph theory according to claim 1 is characterized in that: The carbon monitoring method or carbon accounting method in S2 to obtain the carbon emissions corresponding to each working condition specifically includes the following steps: S21: Determine whether the enterprise has installed carbon monitoring equipment. If yes, proceed to S22; otherwise, proceed to S23. S22. Use carbon monitoring methods to obtain carbon emissions corresponding to energy consumption in each operating condition, i.e., use the company's carbon monitoring equipment to obtain carbon emissions corresponding to energy consumption in each operating condition and process; S23. Use the carbon accounting algorithm to obtain the carbon emissions corresponding to the energy consumption of each operating condition based on energy and power data, that is, use the carbon emission factor method to calculate the carbon emissions corresponding to the energy consumption of each operating condition and process.

4. The industrial production operation optimization method based on graph theory according to claim 1 is characterized in that: The S3 specifically includes the following steps: S31. The independent variables including various operating parameters and the dependent variables including carbon emissions, energy costs, and combustion efficiency are respectively used as vertices of the relationship graph; S32, and according to the relationship between each vertex, use lines to connect the corresponding vertices to obtain the edge of the relationship graph; At this point, we get the relationship diagram between the independent variable and the dependent variable.

5. The industrial production operation optimization method based on graph theory according to claim 1 is characterized in that: The S4 specifically includes the following steps: S41. First, a set of independent variables is given, the set of independent variables including selected values ​​of various operating condition parameters; then, each independent variable in the set of independent variables is divided into multiple parts according to the equal-width binning method, and the value range of each independent variable is determined; S42. Select the qth value of the i-th independent variable from the set of independent variables in S41. , and then select the value from the range of the value of the i-th independent variable Nearest neighbor value ; Then calculate the value and adjacent values The weight distance of ； in, Indicates the qth value of the i-th independent variable and adjacent independent variable values The weight distance of S43. Without considering other independent variables in a set of independent variables in S41, obtain the value of the independent variable according to the relationship diagram Match the frequency of each dependent variable and take the value of the independent variable The frequency of each dependent variable is named weight ; Then according to the weight and the weight of the corresponding dependent variable itself Find the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of ； in, Indicates the qth value of the i-th independent variable The kth value of the corresponding jth dependent variable The weight distance of represents the kth value of the jth dependent variable; is the value of the i-th independent variable Match the value of the jth dependent variable The probability matching weight of , j represents the number of dependent variables, and k is the k-th value of dependent variable j; S44, looping S42 and S43 until the weighted distance between the value of each independent variable in the set of independent variables given in S41 and the value of the adjacent independent variable and the weighted distance between the value of each independent variable in the set of independent variables given in S41 and the value of the corresponding dependent variable are obtained; Then, the weighted distances between the value of each independent variable and the value of the adjacent independent variable in the set of independent variables given in S41 are summed to obtain the sum of the weighted distances. , the calculation formula is as follows: ； Where p is the number of all independent variables; Sum the weighted distances between the value of each independent variable and the value of the corresponding dependent variable in a set of independent variables given in S41 to obtain the sum of the weighted distances , the calculation formula is as follows: ； S45, then based on the sum of weighted distances Sum of weighted distances Construct the objective function; S46. Use simulated annealing algorithm and ant colony algorithm to solve the objective function, optimize the dependent variables, and then obtain the optimal combination of carbon emissions, energy cost and combustion efficiency corresponding to the set of independent variables selected in S41.

6. The method for optimizing industrial production operation based on graph theory according to claim 5, characterized in that: The objective function in S45 is specifically as follows: 。 7. The method for optimizing industrial production operation based on graph theory according to claim 6, characterized in that: The S46 specifically includes the following steps: S461. Using a set number of ants to search for an optimal set of dependent variables corresponding to the set of independent variables selected in S41 in a relationship map between independent variables and dependent variables; each set of dependent variables includes selected values ​​of carbon emissions, energy costs, and combustion efficiency; S462, calculate the probability of ant s solving the next set of dependent variable combinations V; S463. In each iteration, the ant can only modify the value of one unit of the dependent variable in a set of dependent variables in S41. Based on this, the probability of the next set of dependent variable combinations V is converted into a cumulative probability distribution, and the cumulative probability of the dependent variable value being increased or decreased by one unit is obtained. ; The unit is the smallest division unit used in the S41 division process; the dependent variable combination includes three dependent variables, namely carbon emissions, energy cost, and combustion efficiency; S464, according to the roulette algorithm, the roulette wheel is divided into 6 parts, and a random number r is randomly generated. , according to the random number r and the cumulative probability , find satisfaction When the random number lands on the first portion of the roulette wheel, it means the first dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. When the random number lands on the second portion, it means the first dependent variable is adjusted down by one unit, while the other two dependent variables remain unchanged. When the random number lands on the third portion, it means the second dependent variable is adjusted up by one unit, while the other two dependent variables remain unchanged. And so on. The roulette wheel algorithm is used to determine the next optimization direction of each dependent variable. S465, looping through S462 to S464 until a preset iteration stop condition is reached. During each iteration, the pheromones on all paths are volatilized according to a set ratio to achieve pheromone update. After the update, the path with the higher pheromone content is more likely to be selected by the ant; In addition, in each round of iteration, if the new path is better, that is, the difference between the current solution and the historical optimal solution , then the ants will directly search for the optimal solution along the new path, otherwise, they will set the probability Decide whether to solve the optimization along this path; When the number of iterations reaches the set number, the iteration ends, and the optimal carbon emissions, energy costs, and combustion efficiency corresponding to the set of independent variables selected in S41 are obtained.

8. The method for optimizing industrial production operation based on graph theory according to claim 7, characterized in that: The calculation formula for the probability of the next group of ants S solving the dependent variable combination V in S462 is: ； in, and are the parameters set, which are the importance of pheromones and the weight of heuristic information; u, v, and l are the three node positions of the ant, which correspond to the dependent variable combination U, dependent variable combination V, and dependent variable combination L respectively. U is the value combination of this round of iteration, V is the value combination selected for the next round of iteration, and L is any combination in the candidate action set of the ant's next step; represents the dependent variable value combination of ant s at node u, It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination V; It represents the pheromone concentration from the previous dependent variable combination U to the next dependent variable combination L; Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations L, Represents the heuristic function from the previous set of dependent variable combinations U to the next set of dependent variable combinations V, The calculation formula is: ； Where, Indicates the weighted distance of the dependent variable taking the dependent variable combination U; Indicates the weighted distance of the dependent variable from the dependent variable combination V; The cumulative probability in S463 The calculation formula is: ； in, represents action a, i.e. the value of the dependent variable is adjusted up / down by one unit; It represents the corresponding probability of the ant taking any combination of dependent variables in the candidate combinations.

9. The method for optimizing industrial production operation based on graph theory according to claim 8, characterized in that: The manifestation of pheromone volatilization in S465 is: ； in, Indicates the The pheromone of the second iteration; Indicates the The pheromone of the second iteration; represents the pheromone volatilization rate; h is the total number of ants; It means the pheromones released by ants according to their own constructed interpretation; The calculation formula is as follows: ； Where Q is a constant, Indicates the difference between the current solution and the historical optimal solution, , is the path length of ant s; is the historical optimal solution; T is the unit temperature, which decreases with the number of iterations; It is the length of a series of connected sequences formed in a complete iteration process, starting from the starting point and selecting edges or nodes step by step according to the algorithm's path selection rules; The calculation formula of T is as follows: ； in, Indicates the The unit temperature of the iteration; For the The unit temperature of the iteration; is the temperature reduction coefficient.

10. An industrial production operation optimization system based on graph theory, characterized in that: The computer device comprises a computer, wherein the computer device is programmed or configured to execute the industrial production operation optimization method according to any one of claims 1 to 9.

Images