An improved multi-objective genetic algorithm for the optimization problem of a thermoelectric system with thermal energy storage

Abstract

The present invention discloses an improved method for a multi-objective genetic algorithm based on the optimization problem of a thermoelectric system with thermal energy storage, comprising the following steps: analyzing the mechanism model of a dual-extraction unit with a thermal energy storage device, considering the time-series coupling such as the unit ramp constraint; improving the initialization process of the multi-objective genetic algorithm (NSGA-II) in combination with the unit characteristics; improving the crossover process of NSGA-II in combination with the unit characteristics and the improvement of the initialization process; improving the mutation process of NSGA-II in combination with the unit characteristics and the improvement of the initialization process. The embodiment of the present invention improves the search frequency of the algorithm in the feasible region and has a better convergence speed in the search process of the Pareto front.

Description

Technical Field

[0001] The present invention relates to the technical field of load optimal scheduling of a double-extraction heat motor unit with a heat storage device, and particularly to an improved multi-objective genetic algorithm based on an optimization problem of a heat-electricity storage system. Background Art

[0002] Under the background of China's active promotion of energy structure transformation, new energy has developed rapidly in the country. Thermal power plants originally operated in the mode of "determining electricity by heat" in production positioning, that is, only ensuring the heat supply capacity of the unit, and on this basis, responding to the electric load within a limited range. However, in the current situation, higher requirements are placed on its flexibility. Adding a heat storage device to the thermal power unit for flexibility transformation, storing and releasing the excess or insufficient energy inside the system through the energy storage system, and realizing the dynamic balance of energy in each time period. This operation mode of heat-electricity-storage combined regulation strengthens the electric load response ability of the unit in the operation mode of determining electricity by heat and has been widely applied.

[0003] In the optimization process of such systems, the economic function of a cogeneration unit with a heat storage device has the characteristics of being quadratic, non-convex, and multimodal. Conventional optimization methods such as linear programming and Lambda iteration method are difficult to achieve the optimal or near-optimal solution. Therefore, some studies introduce meta-heuristic artificial intelligence optimization algorithms into the process of load optimal scheduling, such as GA, PSO, ABC, etc. In existing studies, the scheduling optimization problem of a thermal power plant with a heat storage device usually only considers the single-objective optimization of economy. Using a multi-objective optimization algorithm to consider both economy and flexibility simultaneously can effectively avoid artificial weighting, and the pareto front obtained by solving can select a strategy that is better from a global perspective for the final solution.

[0004] At the same time, the optimal scheduling of such systems involves the time-sequence complementary characteristics of the energy storage system. The coupling between the power supply and heat supply of the cogeneration unit and the time-sequence coupling characteristics of the energy storage system result in a complex coupling relationship in this type of optimal scheduling problem. During the scheduling process, variable constraints involving time-sequence accumulation within the scheduling interval, such as the upper and lower limits of the output of each device, the ramp constraint, and the charge and discharge power constraint of the energy storage, must also be considered. During the optimization iteration process, when a certain point related to such constraints changes, it may cause the relevant variables at a large number of subsequent moments not to meet the constraints, resulting in extremely low algorithm efficiency. For such constraints, using conventional methods such as penalty functions to handle such complex constraint conditions will cause the algorithm to waste a lot of time searching in the infeasible domain, reducing the search efficiency and making it difficult to achieve the expected optimization effect. Summary of the Invention

[0005] The object of the present invention is to propose an improved multi-objective genetic algorithm for the optimization problem of a thermal power system with heat storage in view of the problems existing in the background technology. The present invention selects the NSGA-II method based on the meta-heuristic algorithm as the optimization method, and improves its initialization and crossover mutation processes in combination with the characteristics of a thermal power plant with heat storage equipment to adapt to complex constraint relationships. The improved algorithm increases the search frequency in the feasible domain, has a good convergence speed, and can approach the Pareto front faster.

[0006] The technical solution of the present invention is an improved multi-objective genetic algorithm for the optimization problem of a thermal power system with heat storage, which is characterized by including the following specific steps:

[0007] S1. Analyze the mechanism model of a dual-extraction unit with heat storage equipment, considering the upper and lower limits of the unit, the heat load equation, and the time-series coupling constraint conditions;

[0008] S2. Improve the initialization process by combining NSGA-II with the characteristics of the unit;

[0009] S3. Improve the crossover process by combining NSGA-II with the characteristics of the unit and the improved initialization process;

[0010] S4. Improve the mutation process by combining NSGA-II with the characteristics of the unit and the improved initialization process.

[0011] Preferably, the heat load equation constraint is as follows:

[0012]

[0013]

[0014] In the formula, is the heat supply of the #1 extraction steam of the #i unit at time t; is the heat supply of the #2 extraction steam of the #i unit at time t; and are the heat load demands of the #1 and #2 extraction steams at time t, respectively; is the heat supply of the heat storage tank to the heat network at time t. When the heat storage tank supplies heat to the outside, it is positive, indicating that the heat in the tank decreases, and vice versa.

[0015] Preferably, the upper and lower limits of the unit are as follows:

[0016]

[0017]

[0018]

[0019] In the formula, and are the changes in the #1 extraction steam heating and #2 extraction steam heating powers of the #i unit at time t; ΔQ i,1max and ΔQ i,2max are the upper limits of the ramping rates of the #1 and #2 extraction steam heating of the #i unit, respectively; is the change in the direct heat supply power of the heat storage tank to the heat network at time t; ΔQ M,Lmax is the upper limit of the change in the heat supply power of the heat storage tank to the heat network.

[0020] Preferably, the specific equation of the heat storage device is as follows:

[0021]

[0022]

[0023] In the formula, h T and h wT are the enthalpy of the heating steam and the enthalpy of the feed water entering the heat storage device, respectively; D Ts is the amount of steam supplied by the heat storage device; S t is the heat capacity in the storage tank at time t; S 0 is the heat output in the storage tank at the initial time; is the heat provided by the hot tank to the heat network through the heat exchanger at time t; Δt is the time interval between two time points.

[0024] Preferably, the improvement of the NSGA-II initialization process in combination with the unit characteristics specifically includes:

[0025] S201. According to the current S 0 , ΔQ M,Lmax , its own upper and lower limit constraints and the upper and lower limit constraints of the heat storage tank storage, the corresponding range constraints can be calculated and obtained ;

[0026] S202. Randomly generate according to the ramping constraint and upper and lower limit constraint of and calculate to judge whether it meets the upper and lower limit constraints and ramping constraints. If it does not meet, re-take and and perform step S202 again; ;

[0027] S203. The current electric power and the obtained in step S202 are known quantities. According to the constraint between the characteristic equation and the double-extraction gas supply amount, the change ranges of and are obtained;

[0028] S204. According to The upper and lower bound constraints and in step S201 the range constraint, and determine whether it can meet the equation constraint. If not, return to step S201 and perform again; if when repeating step S201, the upper bound constraint cannot be obtained, it means that at this time point, due to the existence of the ramping constraint, it cannot be obtained in actual operation, and the time t should be reset and an individual should be regenerated;

[0029] S205. Randomly generate within the constraint range and According to calculate Determine whether it meets the constraint conditions in step S203. If not, reselect and and perform step S205 again;

[0030] S206. Calculate S t＝1 , obtain all the parameters related to the time t = 1, and then iterate the relevant parameters at the time t = 0 to the relevant parameters at the time t = 1, let t = t + 1, and start executing from step 1 again until t = 24.

[0031] Preferably, improve the NSGA-II initialization process in combination with the unit characteristics, and the change range, and the specific equation is as follows:

[0032]

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039]

[0040] Preferably, improve the NSGA-II crossover process in combination with the unit characteristics, specifically including:

[0041] S301. Consider the remaining heat storage capacity S of the hot tank at time t and time t + 1 t '(n) and S t+1(n) range and its own ramping constraint, and give the upper and lower bound constraints of

[0042] S302. Cross the of the parent generation to obtain the offspring. For values exceeding the upper and lower bounds, directly take their limit values, and calculate the heat storage of the hot tank at time t according to ;

[0043] S303. According to and the ramping constraint, given the constraint conditions, cross to obtain the offspring According to calculate to obtain Judge whether it meets the constraint conditions. If not, re - execute step S303;

[0044] S304. According to the constraint conditions of the two - stage extraction steam, obtain the range of change of from Cross to obtain For values exceeding the upper and lower bounds, directly take their limit values, and calculate according to

[0045] S305. According to the compensation logic at time t + 1, at this time and are known quantities; Consider three constraints: ① The upper and lower bound constraints of and generated by the ramping constraint between time t and time t + 2; ② Whether the upper and lower bound constraints of can meet and that is, whether the total electrical load value is between the sum of the maximum values and the sum of the minimum values of the power change ranges of the two units; ;

[0046] S306. Check the data in steps S304 and S305 to judge whether can simultaneously meet its constraint conditions. If not, it means that the value generated at time t in this case cannot meet its own and the constraints of the next moment, and step S301 should be re - executed;

[0047] S307. Cross the parent generation to obtain If it exceeds the constraint, take the extreme value and calculate

[0048] S308. At this time, all variables of the offspring are generated at time t, and the corresponding compensation variables at time t + 1 are also generated accordingly. Iterate the corresponding values at the two times into the parent generation, and give the upper and lower limits of each variable at time t + 1 according to the actual values at time t and the ramp constraint. Let t = t + 1, and continue the next loop, that is, repeat steps S301 to S307 until t = 23.

[0049] Preferably, the NSGA-II crossover process is improved in combination with the unit characteristics. The range of change is as follows:

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057] In the formula, u is a random number between 0 and 1; μ is the set crossover change coefficient. and are the heat supply of the #1 extraction steam of the #i unit corresponding to the offspring 1 and the offspring 2 at time t, respectively. and are the heat supply of the #1 extraction steam of the #i unit corresponding to the parent 1 and the parent 2 at time t, respectively. and are the heat supply of the #1 extraction steam of the #i unit corresponding to the offspring n and the parent n at time t + 1, respectively.

[0058] Preferably, the NSGA-II crossover process is improved in combination with the unit characteristics. and The range of change is as follows:

[0059]

[0060]

[0061]

[0062]

[0063] Preferably, the NSGA-II crossover process is improved in combination with the unit characteristics. The variation range is as follows, and the specific equation is as follows:

[0064]

[0065]

[0066]

[0067] In the formula, the subscripts i and 2 represent the #2 extraction steam heating of the #i unit.

[0068] A system of an improved multi-objective genetic algorithm method based on the optimization problem of a heat storage integrated thermoelectric system is used to implement the above-mentioned optimization algorithm improvement. The system includes:

[0069] An initialization process improvement module for generating an initial population that satisfies the upper and lower limit constraints of the unit, the ramp constraint, and the time series coupling constraint of the heat storage device;

[0070] A crossover process improvement module for performing a crossover operation based on the initialized population that satisfies the constraints generated in the initialization process, and obtaining a subpopulation that satisfies the constraint conditions through joint optimization of the previous and next two times during the iteration process;

[0071] A mutation process improvement module for performing a mutation operation based on the initialized population that satisfies the constraints generated in the initialization process, and obtaining a subpopulation that satisfies the constraint conditions through joint optimization of the previous and next two times during the iteration process.

[0072] A terminal includes a processor and a storage medium;

[0073] The storage medium is used to store instructions;

[0074] The processor is used to operate according to the instructions to execute the steps of the above method.

[0075] A computer-readable storage medium stores a computer program, and when the program is executed by a processor, it implements the steps of the above method.

[0076] Compared with the prior art, the present invention has the following beneficial technical effects:

[0077] In the present invention, the value of the constraint equation of the variable containing the ramp constraint is defined as the state. When an individual gene segment changes under the influence of crossover and mutation at a certain moment, a new constraint is imposed on the subsequent segment through the state library of the constraint state, minimizing the change of the variable constraint at the subsequent moment. In this process, the influence of the parent generation on the offspring is taken into account, and at the same time, it can ensure that the newly generated offspring meet all the constraint conditions. When generating the individual gene sequence, the ramp constraint is considered sequentially according to the time series to ensure that all gene segments of the individual meet the constraint conditions. During the crossover and mutation iteration process, when a variable changes at a certain moment, the relevant variables of the time series coupling constraint at the next moment are changed to ensure that all subsequent moments are within the feasible region. The improved NSGA-II algorithm can ensure that the modification of an individual gene segment during the optimization process will not cause its offspring to exceed the feasible region, increasing the frequency of searching within the feasible region. Compared with the original method, it has obvious advantages in the search efficiency on the Pareto front. Description of the Drawings

[0078] Figure 1 is a schematic flow chart of an improved multi-objective genetic algorithm for the optimization problem of a thermoelectric system with thermal energy storage according to an embodiment of the present invention;

[0079] Figure 2 is a schematic flow chart of an improved NSGA-II initialization process combined with unit characteristics according to an embodiment of the present invention;

[0080] Figure 3 is a schematic flow chart of an improved NSGA-II crossover process combined with unit characteristics and an improved initialization process according to an embodiment of the present invention;

[0081] Figure 4 is a schematic diagram of a thermoelectric system with thermal energy storage according to an embodiment of the present invention;

[0082] Figure 5 is a schematic diagram of the thermoelectric coupling relationship of a thermal power unit according to an embodiment of the present invention;

[0083] Figure 6 is a schematic diagram comparing the Pareto fronts of the two algorithms after 500 iterations according to an embodiment of the present invention;

[0084] Figure 7 is a schematic diagram comparing the Pareto fronts of the two algorithms after 5000 iterations according to an embodiment of the present invention. Detailed Embodiments

[0085] Embodiment 1

[0086] As Figure 4 shown, Embodiment 1 of the present invention provides an improved multi-objective genetic algorithm for the optimization problem of a thermoelectric system with thermal energy storage. In a preferred but non-limiting embodiment of the present invention, the method includes the following steps:

[0087] S1. Analyze the mechanism model of the double-extraction unit with a heat storage device, considering the upper and lower limits constraints of the unit, the heat load equation, and the time-series coupling constraints;

[0088] S2. Improve the NSGA-II initialization process by combining the characteristics of the unit;

[0089] S3. Improve the NSGA-II crossover process by combining the characteristics of the unit and the improved initialization process;

[0090] S4. Improve the NSGA-II mutation process by combining the characteristics of the unit and the improved initialization process.

[0091] The parameters include the parameters of Unit A, the parameters of Unit B, and the parameters of the heat storage device, which are specifically as follows:

[0092] Unit A:

[0093] h 0,A = 3500 kJ / kg, h e,A = 2800 kJ / kg, h wf,A = 1300 kJ / kg, η b,A = 0.96, c v1,A = c v2,A = -0.13, c m,A = 0.76, k 1,A = 0.3292, k 2,A = -0.1655, k 3,A = 9.3681, a A = 1200, μ 1,A = 600, μ 2,A = 800

[0094] Unit B:

[0095] h 0,B = 3600 kJ / kg, h e,B = 3000 kJ / kg, h wf,B = 1500 kJ / kg, η b,B = 0.98, c v1,B = c v2,B = -0.15, c m,B = 0.85, k 1,B = 0.3742, k 2,B = -0.1437, k 3,B = 11.4258, a B = 1200, μ 1,B = 600, μ 2,B = 800

[0096] Heat storage equipment:

[0097] S max = 200 MWh, ΔQ M,Lmax = 80 MW

[0098] In this example, the unit is a single-extraction non-reheat unit, and a heat storage equipment is equipped for its extraction steam for heating. In the model of this example, there is no energy storage for the electric load, and it is necessary to always meet the demand of the power grid load and there are ramp constraints. There is a thermoelectric coupling characteristic between the heat load and the electric load. As Figure 5 shown, that is, when the electric load is determined, the heat load can only vary within a certain range. Since the unit has no reheat and the main steam enthalpy and feed water enthalpy do not change, the coal consumption of the unit is only related to the main steam flow:

[0099]

[0100] c m = ΔP A / ΔQ A is the slope of the electric power and heat power during backpressure operation, also known as the elasticity coefficient, which is usually considered a constant. In the formula, ΔP A is the change in electric power during the backpressure operation (BC section) of the unit, and ΔQ A is the change in heat power during the backpressure operation (BC section) of the unit; c v is the decrease in electric power per unit of extracted heating heat when the steam inlet flow remains unchanged, that is, the slope of the AB section. Among them, c v1 is the corresponding value of c v at the maximum electric power output, and c v2 is the corresponding value of c v at the minimum electric power output; is the heating power corresponding to the minimum power generation (backpressure operation); is the maximum heating output of the unit (backpressure operation); and respectively represent the minimum and maximum electric powers (active power output) of the unit under the pure condensing condition.

[0101] According to the electrothermal characteristics of the unit, within the allowable range of heating, the lower and upper limits of the electric power are respectively:

[0102]

[0103]

[0104] In the formula, is the lower limit of the power generation of the i-th unit at time t, MW; are respectively the minimum power generations of the i-th unit under the pure condensing condition and the pure backpressure condition, MW; is the heating power of the i-th unit at time t, MW; They are the minimum and maximum heating powers of the i-th unit under the pure condensing condition, in MW.

[0105] The targeted improvement of the algorithm in this paper is mainly based on the determination of the electrical load, considering that the two extractions each meet the load demand and there is a coupling relationship among the three. At the same time, a heat storage device is configured for the extraction #2 for heating. In this example, the electrical load corresponds to the extraction #1 in the model, and the heating load corresponds to the extraction #2 in the model. Only the coupling relationship between the two is simplified according to the previous method, and they essentially belong to the same type of optimization problem.

[0106] The two selected objective functions are the comprehensive operating cost objective function and the peak shaving capacity objective function. Taking the comprehensive operating cost of the system as the objective function F1, which includes the operating cost of the combined heat and power unit and the economic compensation for participating in peak shaving, the specific objective function is as follows:

[0107]

[0108]

[0109]

[0110] In the formula, F1 is the comprehensive operating cost of the system; a i is the comprehensive coal price coefficient of the i-th unit, is the coal consumption of the i-th unit at the t-th moment, in t; P i,50％ and P i,40％ are the electrical loads corresponding to 50% and 40% loads of the i-th unit, in MW; is the electrical load of the i-th unit at the t-th moment, in MW; t 1,i and t 2,i are all the time nodes of the i-th unit between 50% and 40% loads and below 40% load respectively; Δt is the time difference between adjacent time nodes, in h; μ1 and μ2 are the quotes for the 40% and 30% load levels respectively, in yuan / (kW·h).

[0111] To improve the load curve and enhance the peak shaving capacity of the unit during operation, considering the load change of the unit itself and the heat storage in the heat storage tank, the comprehensive peak shaving capacity objective function is defined as:

[0112]

[0113]

[0114]

[0115] In the formula, F2 is the comprehensive peak shaving capacity of the system; T1 and T2 represent all the time nodes during the low electricity consumption valley and the high electricity consumption peak respectively.

[0116] Using the parameters of this example and combining the improvements made to the initialization and crossover mutation processes in this paper, the running results of the following three algorithms are compared: the original NSGA-II algorithm, the NSGA-II algorithm with only the initialization process modified, and the NSGA-II algorithm with both the initialization and crossover processes modified. The original NSGA-II method is that the initialized individuals are randomly valued within the upper and lower limits of the independent variable. In the crossover and mutation processes, all gene segments of the individuals do not consider the time series coupling and are operated in the order from front to back. The change amount during the operation is only related to the set crossover coefficient and mutation coefficient, and the penalty function method is used to punish the individuals that do not meet the constraint conditions. The improved initialization method is to process the initial population using the improved initialization method, and the subsequent crossover and mutation processes are the same as the original method, and the penalty function method is used to handle the individuals that do not meet the constraints. The initialization / crossover improvement method uses the improved initialization and crossover mutation algorithms and does not use the penalty function method for processing.

[0117] The parameter settings of the example are as follows: The maximum number of iterations is set to 5000 and 500 respectively, the population size is 100, the single-precision real number coding method is used, the double-person tournament rule is adopted for the selection of dominant individuals, the crossover of individuals is to randomly select two parent individuals for crossover, the crossover probability is 0.8, and the mutation probability is 0.2.

[0118] The test platform configuration is as follows: CPU: Intel Core i7-10875H 2.300 MHz (4.587 MHz); GPU: NVIDIA GeForce RTX 2070 Super with Max-Q Design (8192 MB) (930 MHz); Programming environment: Matlab R2021a

[0119] Each group is run 10 times to obtain the average running time. From the perspective of time, when the number of iterations is 5000 times, the cpu times of the three algorithms are 144.58, 264.90, and 470.51 respectively. The improved method does not have an advantage in running speed. Among them, the improvement of the crossover mutation has a greater impact on the calculation time of the algorithm, doubling the time basically. When the number of iterations is 500 times, the time difference between the three methods decreases significantly, which are 17.12, 25.42, and 28.60 respectively. The relative change in time between the initialization / crossover improvement method and the original algorithm decreases from 225.43% before to 67.06%. After the number of iterations is reduced by 10 times, the time consumed by the original calculation method is 11.84% of the original, while the improved method is only 6.08% of the original. It is not difficult to see that in the case of a small number of iterations, the disadvantage of the initialization / crossover improvement method in running time is not obvious, but this disadvantage will increase with the increase of the number of iterations.

[0120] The reason is that the original algorithm only considers the upper and lower bound constraints and assigns penalties to its objective function when other constraint conditions are not met. Therefore, the computational work performed in each iteration is almost the same and will not vary due to differences in individuals or constraint conditions. The improved algorithm requires that each individual satisfy all constraint conditions during initialization and execution. Since the time series constraints are given priority during generation, this actually compresses the range within which variables can vary. Therefore, when faced with thermoelectric coupling constraints, a large number of infeasible solutions will be generated. Before feasible solutions are generated, the current iteration will not end but will continue to search for new solutions until feasible solutions are found, which will consume additional time. As the number of iterations increases and the solution set approaches the Pareto front, the genes of the parent individuals will become more stable, which makes the search range of gene segments become narrower. Therefore, the search for feasible solutions will become more difficult and require more time.

[0121] Typical results of the initialization / crossover improvement method and the initialization improvement method are selected to plot their Pareto front diagrams. The results of 5000 iterations and 500 iterations are shown respectively as Figure 6 and Figure 7 shown. Since the original algorithm did not converge under both iteration conditions, all individuals did not satisfy the constraint conditions, and its objective function was greatly affected by the penalty function, so it will not be discussed in detail.

[0122] From the perspective of the Pareto front, the Pareto front of the initialization / crossover improvement method has obvious advantages compared with the other two methods. In both 500 iterations and 5000 iterations, the Pareto front of the initialization / crossover improvement method is located below and to the left of the other two, indicating that its solution set dominates the solution sets of the other two methods under the same number of iterations. The initialization improvement method makes the optimization direction and effectiveness of the results of NSGA-II far exceed those of the original NSGA-II method. The fundamental reason is that the original algorithm did not operate reasonably during the initialization process, resulting in individuals generated during initialization being unable to meet the relatively complex time constraint conditions. At the same time, the time series constraint conditions were not processed during the subsequent change process, and the generated offspring individuals basically could not meet the constraint conditions, and were greatly affected by the penalty function in the case of fewer iterations; on the basis of the former, the initialization / crossover improvement method ensures that all individuals generated by the subsequent crossover and mutation operations can maintain the advantages of the initialization stage, that is, all individuals generated from the initialization stage to the optimization stage can meet the requirements of the constraint conditions. Therefore, under the same number of iterations, its Pareto front is better.

[0123] The invention discloses an improved multi-objective genetic algorithm method based on the optimization problem of a thermal power system containing heat storage. The constraint equation value of the variable containing the climbing constraint is defined as a state. When the individual gene fragment changes under the influence of crossover and mutation at a certain moment, a new constraint is generated for the subsequent fragment through the state library of the constraint state, so that the variable constraint change at the subsequent moment is minimized. Specifically, in the iterative process, when a variable is crossed or mutated at time t to make it change, the variable at time t+1 is changed synchronously to offset the change of the variable at time t as much as possible, so that at the end of time t+1, the various constraints of the input at time t+2 maintain the minimum change, thereby ensuring that the subsequent optimization variables remain stable and meet various constraints, and iterate all the generated variables at time t and the partially compensated variables at time t+1 to the corresponding moment of the parent generation. After the generation of the parameters at time t is completed, similar operations are performed on the variables at time t+1, and the variables at time t+2 are used to compensate for the changes at time t+1, so as to ensure that the various constraints at time t+3 are kept constant. In this process, the influence of the parent generation on the offspring is taken into account, and at the same time, it can ensure that the generated new offspring meets all constraints. When generating individual gene sequences, the climbing constraints are considered in accordance with the time series to ensure that all gene fragments of the individual meet the constraints. In the process of crossover mutation iteration, when the variables change at a certain moment, the variables related to the time series coupling constraints at the next moment are changed to ensure that all subsequent moments are within the feasible domain. The improved NSGA-II algorithm can ensure that the modification of an individual gene fragment during the optimization process will not cause its offspring to exceed the feasible domain, increase the frequency of the algorithm's search within the feasible domain, and improve the efficiency of the algorithm's search.

[0124] The embodiment of the present invention provides a multi-objective genetic algorithm improvement method system based on the optimization problem of a thermal power system with heat storage, which is used to implement the optimization method. The system includes:

[0125] The initialization process improvement module is used to generate the initial population that meets the upper and lower limit constraints of the unit, the ramp constraint, and the timing coupling constraint of the heat storage equipment;

[0126] The crossover process improvement module is used to perform a crossover operation based on the initialization population that meets the constraints generated by the initialization process, and obtain a child population that meets the constraints by jointly optimizing the previous and next times during the iteration process;

[0127] The mutation process improvement module is used to perform mutation operations based on the initialization population that meets the constraints generated by the initialization process, and obtain the offspring population that meets the constraints by jointly optimizing the previous and next times during the iteration process.

[0128] A terminal includes a processor and a storage medium; the storage medium is used to store instructions;

[0129] The processor is configured to operate in accordance with the instructions to perform the steps of the method.

[0130] A computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements the steps of the method.

[0131] The present disclosure may be a system, a method, and / or a computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement aspects of the present disclosure.

[0132] A computer-readable storage medium may be a tangible device that can retain and store instructions for use by an instruction execution device. A computer-readable storage medium may be, for example—but not limited to—an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punch card or raised structures in a groove having instructions stored thereon, and any suitable combination of the foregoing. The computer-readable storage medium as used herein is not to be construed as a transitory signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., an optical pulse through an optical fiber cable), or an electrical signal transmitted through a wire.

[0133] The computer-readable program instructions described herein may be downloaded to respective computing / processing devices from a computer-readable storage medium or may be downloaded to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network, and / or a wireless network. The network may include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0134] Computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine - related instructions, microcode, firmware instructions, state - setting data, or source code or object code written in any combination of one or more programming languages, including object - oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer - readable program instructions may be executed entirely on the user's computer, partially on the user's computer, executed as a stand - alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it may be connected to an external computer (e.g., via an Internet service provider through the Internet). In some embodiments, by using the state information of computer - readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field - programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer - readable program instructions to implement various aspects of the present disclosure.

[0135] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those skilled in the art to which the present invention pertains.

Claims

1. An improved multi-objective genetic algorithm for the optimization problem of a thermoelectric system with heat storage, characterized in that, It includes the following specific steps: S1. Analyze the mechanism model of the double-extraction unit with a heat storage device, considering the upper and lower limits constraints of the unit, the heat load equation, and the time-series coupling constraints; S2. Improve the initialization process of NSGA-II combined with the characteristics of the unit; S3. Improve the crossover process of NSGA-II combined with the characteristics of the unit and the improved initialization process; S301. Consider the remaining heat storage quantity S of the hot tank at time t and time t+1 t '(n) and S t+1 (n) range and its own ramping constraint, and give upper and lower limit constraints; S302. Cross the of the parent generation to obtain offspring. For values exceeding the upper and lower limits, directly take their limit values, and calculate the heat storage of the hot tank at time t according to ; S303. According to and the ramp-up constraint, given the constraint conditions of , perform crossover on to obtain the offspring According to , calculate to obtain Judge whether it meets the constraint conditions. If not, repeat step S303; S304. According to the constraint conditions of the two extractions, from obtain the variation range of, and from cross to obtain For values exceeding the upper and lower limits, directly take their limit values, and according to calculate S305: According to the compensation logic at time t+1, and is a known quantity; consider three constraints: ① at time t and t+2 and The resulting climbing constraint The upper and lower limit constraints; ② and Can the upper and lower limit constraints of That is, whether the total electric load value is between the sum of the maximum value and the sum of the minimum value of the power variation range of the two units; S306. Check the data in steps S304 and S305 to determine whether the constraint conditions can be satisfied simultaneously. If not, it means that the value generated at time t in this case cannot meet its own and the constraints of the next moment, and step S301 should be executed again; S307. Obtain according to the parent generation by crossover If it exceeds the constraint, take the extreme value and calculate using the heat load equation constraint S308. At this time, all variables of the offspring at time t are generated, and the corresponding compensation variables at time t + 1 are also generated correspondingly. Iterate the corresponding values of the two times into the parent generation, and give the upper and lower limits of each variable at time t + 1 according to the actual value and ramp constraint at time t. Let t = t + 1, and continue the next cycle, that is, repeat steps S301 to S307 until t = 23; S4. Improve the mutation process of NSGA-II combined with the characteristics of the unit and the improved initialization process.

2. The improved multi-objective genetic algorithm method based on the optimization problem of a thermal energy storage integrated thermoelectric system according to claim 1, wherein The heat load equation constraint, the specific equation is as follows: Wherein, is the heat supply of the #1 extraction steam of the #i unit at time t; is the heat supply of the #2 extraction steam of the #i unit at time t; and are the heat load demands of the #1 and #2 extraction steams at time t, respectively; is the heat supply of the heat storage tank to the heat network at time t. When the heat storage tank supplies heat to the outside, it is positive, indicating a decrease in the heat in the heat tank, and vice versa.

3. The improved multi-objective genetic algorithm method based on the optimization problem of a heat storage thermoelectric system according to claim 2, wherein The upper and lower limits constraint of the unit, the specific equation is as follows: wherein, and are respectively the changes in the heating powers of the #1 extraction steam and the #2 extraction steam of the #i unit at time t; ΔQ i,1max and ΔQ i,2max are respectively the upper limits of the ramping rates of the #1 and #2 extraction steam heating of the #i unit; is the change in the direct heat supply power of the heat network by the heat storage tank at time t; ΔQ M,Lmax is the upper limit of the change in the heat supply power of the heat storage tank to the heat network.

4. The improved multi-objective genetic algorithm method based on the optimization problem of a thermal energy storage integrated thermoelectric system according to claim 3, wherein, The heat storage device, the specific equation is as follows: where h T and h wT are the enthalpy of the heating steam and the feed water entering the heat storage device respectively; D Ts is the amount of steam supplied by the heat storage device; S t is the heat capacity in the storage tank at time t; S 0 is the heat output in the storage tank at the initial time; is the heat provided by the hot tank to the heat network through the heat exchanger at time t; Δt is the time interval between two time points.

5. The improved multi-objective genetic algorithm method based on the optimization problem of a heat storage thermoelectric system according to claim 4, wherein The improvement of the NSGA-II initialization process combined with the characteristics of the unit specifically includes: S201. According to the current S 0 , ΔQ M,Lmax , its own upper and lower limit constraints and the upper and lower limit constraints of the heat storage tank storage capacity, the corresponding range constraints can be calculated and obtained ; S202. Randomly generate according to the ramp constraint and upper and lower limit constraints of and calculate according to Determine whether it meets the upper and lower limit constraints and ramp constraints. If not, re-obtain and and repeat step S202; and ​ S203, current electric power and the one obtained in step S202 are known quantities. According to the characteristic equation, the constraints between the generated and the double-extraction gas supply volume are obtained and range of variation; S204. According to the upper and lower bound constraints and in step S201 the range constraints, determine whether it can satisfy the equality constraint. If not, return to step S201 and repeat; if when repeating step S201, the upper bound constraint cannot be obtained, it means that at this time point, due to the existence of the ramp constraint, it cannot be obtained in actual operation, and the time t should be reset and an individual should be regenerated. S205. Randomly generate within the constraints and According to Calculate Judge whether it meets the constraint conditions in step S203. If not, re - obtain and And perform step S205 again; S206. Calculate S t＝1 , obtain all the parameters related to the moment of t = 1. Then, iterate the relevant parameters at the moment of t = 0 to the relevant parameters at the moment of t = 1, let t = t + 1, and start over from step 1 until the moment of t = 24.

6. The improved multi-objective genetic algorithm method based on the optimization problem of a heat storage thermoelectric system according to claim 5, wherein Improvement of the NSGA-II Initialization Process Combining Unit Characteristics and The range of variation is as follows: The specific equation is as follows:

7. The improved multi-objective genetic algorithm method based on the optimization problem of a thermal energy storage integrated thermoelectric system according to claim 6, wherein Improvement of NSGA-II Crossover Process Combining Unit Characteristics The range of variation is as follows, and the specific equation is as follows: where \(u\) is a random number between 0 and 1; \(\mu\) is the set cross-variation coefficient; and are the heat supply amounts of the first extraction steam of the \(i\)-th unit corresponding to the first offspring and the second offspring at time \(t\), respectively; and are the heat supply amounts of the first extraction steam of the \(i\)-th unit corresponding to the first parent and the second parent at time \(t\), respectively; and are the heat supply amounts of the first extraction steam of the \(i\)-th unit corresponding to the \(n\)-th offspring and the \(n\)-th parent at time \(t + 1\), respectively.

8. The improved multi-objective genetic algorithm method based on the optimization problem of a thermal energy storage integrated thermoelectric system according to claim 7, wherein Improvement of NSGA-II Crossover Process Combining Unit Characteristics and The range of variation is as follows: The specific equation is as follows:

9. The improved multi-objective genetic algorithm method based on the optimization problem of a heat storage thermoelectric system according to claim 8, characterized in that, Improvement of the NSGA-II Crossover Process Combining Unit Characteristics The range of variation is as follows. The specific equation is as follows: In the formula, the subscript i, 2 represents the #2 extraction steam heating of the #i unit.

Images