Fire storage system operation strategy optimization method based on multi-objective evaluation method
Through the multi-objective evaluation method and the multi-objective gray wolf algorithm, the coupled molten salt heat storage system of coal-fired units is solved, and the problems of low thermal efficiency and insufficient peak-shaving performance are achieved, the system's heat consumption rate and peak-shaving capacity are improved. Energy efficiency and flexibility have promoted the sustainable development of the energy industry.
Patent Information
- Application Number
- CN202411031320.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-15
AI Technical Summary
The thermal efficiency of the coal-fired unit is low in heat storage process of coupled molten salt and insufficient peak regulating performance, so the system operation strategy needs to be optimized to improve energy efficiency and flexibility.
Using multi-objective evaluation method, establish a system optimization operation strategy model, set the system heat consumption rate and peak shaving capacity as dual-objective functions, optimize it using multi-objective gray wolf algorithm, and introduce multi-dimensional preference analysis method and penalty function method to solve the constraints Problem, optimize the unit output power and heat consumption rate.
It has achieved the minimization of system heat consumption rate and maximized peak shaving capacity, improved energy efficiency and system flexibility, ensured the feasibility and practicality of optimization results, and promoted the sustainable development of the energy industry.
Smart Images

Figure CN118966672A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of peak regulation of thermal power generation technology, and in particular to a method for optimizing operation strategy of a thermal storage system based on a multi-objective evaluation method. Background Art
[0002] The molten salt thermal storage unit stores the heat of the excess high-temperature steam generated by the boiler by adding a molten salt thermal storage device between the boiler and the steam turbine, and returns the heat to the steam turbine through the molten salt energy release system according to the grid load requirements, thereby realizing the time migration of power and energy, making the rigidly connected wind, solar, thermal and storage integrated system flexible, thereby absorbing more renewable energy. However, due to the conversion of high-grade thermal energy to low-grade thermal energy and steam throttling in the process of coupling molten salt thermal storage, the thermal efficiency of coal-fired units is usually lower than that of conventional coal-fired units. Therefore, it is necessary to comprehensively consider the economy and peak-shaving performance of the coupled system to optimize the system operation strategy, so as to achieve efficient and coordinated operation of the thermal storage system. Summary of the invention
[0003] The present invention is based on the operation mechanism of the coal-fired unit coupled heat storage system, establishes a system optimization operation strategy model based on a multi-objective evaluation method, sets the system heat rate and peak load capacity as a dual-objective function, and considers the unit constraints. Then, the multi-objective grey wolf algorithm is used for optimization and solution, and a set of solutions with the best comprehensive performance is output.
[0004] In order to achieve the above purpose, the present invention adopts the following technical scheme:
[0005] The optimization goal of the operation strategy of the present invention is to achieve the comprehensive optimization of the system heat rate and peak load capacity at the same time, that is, under the given boundary conditions of the heat storage process system operation, the system heat rate HC is minimized and the system peak load capacity ΔW is minimized. ch During the optimization process, the peak load capacity ΔW ch Converted into unit output power, the above objectives can be transformed into achieving the lowest system output power and minimizing the system heat rate. The objective function is shown in formula (1).
[0006]
[0007] Where: HC is the objective function of system heat rate;
[0008] W ch is the objective function of the system peak load capacity;
[0009] T HT , T HTmin and T HTmax is the heat storage temperature parameter, the minimum and maximum values of the heat storage temperature, K;
[0010] and is the thermal storage steam extraction flow parameter, the minimum and maximum thermal storage steam extraction flow, t / h;
[0011] β, β min and β max is the split ratio parameter, the minimum and maximum split ratio;
[0012] is an inequality constraint;
[0013] is an equality constraint.
[0014] In multi-objective optimization problems, two objectives are in direct competition with each other. Therefore, the selection of the optimal compromise solution plays a vital role in multi-objective optimization problems. For this reason, the multidimensional preference analysis method is introduced to screen the optimal compromise solution. This method introduces the ideal point y * The concept of ideal point is the point composed of the optimal solutions of each single objective. This point is not located on the Pareto frontier and is an unattainable ideal point, as shown in formula (2).
[0015] y * =(y1 * ,y2 * ) (2)
[0016] Where: y1 * is the optimal function value considering objective function 1;
[0017] y2 * is the optimal function value considering objective function 2.
[0018] The multidimensional preference analysis method calculates the difference between each non-dominated solution on the Pareto front and the ideal point y * The Euclidean distance between the two points is used to select the point y′ with the shortest distance to the ideal point as the compromise optimal solution. The distance between the ideal point and any non-dominated solution in the solution set is the Euclidean distance, and the calculation formula is shown in formula (3).
[0019]
[0020] In the formula: y, y * are non-dominated solutions and ideal points respectively;
[0021] y1,y1 * The system heat rate function values for the non-dominated solution and the ideal optimal solution;
[0022] y2,y2 * is the output power function value of the non-dominated solution and the ideal optimal solution.
[0023] Find the ideal point y in the non-dominated solution set *The point with the shortest Euclidean distance is taken as the compromise optimal solution y′, which can be calculated using formula (4).
[0024]
[0025] The constraint violation degree when individual x violates the equality constraint is shown in formula (5), and the constraint violation degree when violating the inequality constraint is shown in formula (6).
[0026]
[0027] The total violation degree of individual x is shown in formula (7).
[0028]
[0029] The penalty function method is used to solve the constraint problem in the optimization model of this paper. The unit constraint conditions are converted into penalty terms and added to the objective function. Then the constrained optimization problem shown in formula (1) is transformed into an unconstrained optimization problem shown in formula (8).
[0030]
[0031] Multi-objective Grey Wolf Algorithm
[0032] First, all individuals are sorted according to the fitness function. The optimal solution corresponds to the α wolf in the wolf pack. The two suboptimal solutions are selected as β wolf and δ wolf. The remaining feasible solutions are ω wolf. The ω individual is optimized with the first three level individuals as the guide. Then, the hunting behavior of the wolf pack is imitated using formula (9) and formula (10). Each ω individual uses the update mechanism to generate the next generation solution according to the position of the three optimal solutions.
[0033]
[0034]
[0035] Where: is the prey position vector;
[0036] is the hunter position vector;
[0037] is the coefficient vector.
[0038] The hunter generates a new position vector according to the prey's position and the random coefficient, as shown in equations (11) and (12).
[0039]
[0040] Where: is a random parameter, taking a random number from [0,1] each time it is calculated;
[0041] is the global convergence factor, which decreases from 2 to 0 during the iteration process.
[0042] In the solution process, it is necessary to estimate the optimal solution based on α, β, and δ individuals. The position of
[0043] and (14).
[0044]
[0045] Finally, the three updated positions are averaged using formula (15) as the result of this round of update.
[0046]
[0047] The multi-objective gray wolf algorithm has made two changes based on the gray wolf algorithm. The first is to introduce the concept of archive, which stores the non-dominated solutions generated in each iteration process. The second is to use the roulette strategy to select the top three gray wolves from the archive, and the probability of each individual being selected is P i It is inversely proportional to the number of individuals in the Archive, as shown in formula (16).
[0048] P i =c / N i (16)
[0049] Where: c is a constant; N i is the total number of individuals in the group to which the individual belongs.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] This method can simultaneously optimize energy efficiency and power system flexibility by setting the system heat rate and peak load capacity as dual objective functions. In addition, this method fully considers the unit constraints in the heat storage process in the optimization model, including but not limited to energy balance, temperature limit, pressure limit, etc., to ensure the feasibility and practicality of the optimization results. This method of comprehensive consideration of constraints avoids the problem that the theoretical optimal solution cannot be achieved in actual operation. The research and application of this method can not only improve the performance of existing energy systems, but also stimulate technological innovations in related fields, such as more efficient molten salt heat storage technology, smarter control systems, etc., thereby promoting the sustainable development of the energy industry. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0053] The present invention is further described below in conjunction with the accompanying drawings.
[0054] The optimization goal of the operation strategy of the present invention is to achieve the comprehensive optimization of the system heat rate and peak load capacity at the same time, that is, under the given boundary conditions of the heat storage process system operation, the system heat rate HC is minimized and the system peak load capacity ΔW is minimized. ch During the optimization process, the peak load capacity ΔW ch Converted into unit output power, the above objectives can be converted into achieving the lowest system output power and minimizing the system heat rate. The objective function is as follows:
[0055] shown.
[0056]
[0057]
[0058] Where: HC is the objective function of system heat rate;
[0059] W ch is the objective function of the system peak load capacity;
[0060] T HT , T HTmin and T HTmax is the heat storage temperature parameter, the minimum and maximum values of the heat storage temperature, K;
[0061] and is the thermal storage steam extraction flow parameter, the minimum and maximum thermal storage steam extraction flow, t / h;
[0062] β, β min and β max is the split ratio parameter, the minimum and maximum split ratio;
[0063] is an inequality constraint;
[0064] is an equality constraint.
[0065] In multi-objective optimization problems, two objectives are in direct competition with each other. Therefore, the selection of the optimal compromise solution plays a vital role in multi-objective optimization problems. For this reason, the multidimensional preference analysis method is introduced to screen the optimal compromise solution. This method introduces the ideal point y * The concept of ideal point is the point composed of the optimal solutions of each single objective. This point is not located on the Pareto frontier and is an unattainable ideal point, as shown in formula (2).
[0066] y * =(y1 * ,y2 * ) (2)
[0067] Where: y1 * is the optimal function value considering objective function 1;
[0068] y2 * is the optimal function value considering objective function 2.
[0069] The multidimensional preference analysis method calculates the difference between each non-dominated solution on the Pareto front and the ideal point y * The Euclidean distance between the two points is used to select the point y′ with the shortest distance to the ideal point as the compromise optimal solution. The distance between the ideal point and any non-dominated solution in the solution set is the Euclidean distance, and the calculation formula is shown in formula (3).
[0070]
[0071] In the formula: y, y * are non-dominated solutions and ideal points respectively;
[0072] y1,y1 * The system heat rate function values for the non-dominated solution and the ideal optimal solution;
[0073] y2,y2 * is the output power function value of the non-dominated solution and the ideal optimal solution.
[0074] Find the ideal point y in the non-dominated solution set * The point with the shortest Euclidean distance is taken as the compromise optimal solution y′, which can be calculated using formula (4).
[0075]
[0076] The constraint violation degree when individual x violates the equality constraint is shown in formula (5), and the constraint violation degree when violating the inequality constraint is shown in formula (6).
[0077]
[0078] The total violation degree of individual x is shown in formula (7).
[0079]
[0080] The penalty function method is used to solve the constraint problem in the optimization model of this paper. The unit constraint conditions are converted into penalty terms and added to the objective function. Then the constrained optimization problem shown in formula (1) is transformed into an unconstrained optimization problem shown in formula (8).
[0081]
[0082] Multi-objective Grey Wolf Algorithm
[0083] First, all individuals are sorted according to the fitness function. The optimal solution corresponds to the α wolf in the wolf pack. The two suboptimal solutions are selected as β wolf and δ wolf. The remaining feasible solutions are ω wolf. The ω individual is optimized with the first three level individuals as the guide. Then, the hunting behavior of the wolf pack is imitated using formula (9) and formula (10). Each ω individual uses the update mechanism to generate the next generation solution according to the position of the three optimal solutions.
[0084]
[0085]
[0086] Where: is the prey position vector;
[0087] is the hunter position vector;
[0088] is the coefficient vector.
[0089] The hunter generates a new position vector based on the prey position and the random coefficient, as shown in formula (11):
[0090] And shown in formula (12).
[0091]
[0092]
[0093] Where: is a random parameter, taking a random number from [0,1] each time it is calculated;
[0094] is the global convergence factor, which decreases from 2 to 0 during the iteration process.
[0095] In the solution process, it is necessary to estimate the optimal solution based on α, β, and δ individuals. The position of
[0096] and (14).
[0097]
[0098] Finally, the three updated positions are averaged using formula (15) as the result of this round of update.
[0099]
[0100] The multi-objective gray wolf algorithm has made two changes based on the gray wolf algorithm. The first is to introduce the concept of archive, which stores the non-dominated solutions generated in each iteration process. The second is to use the roulette strategy to select the top three gray wolves from the archive, and the probability of each individual being selected is P i It is inversely proportional to the number of individuals in the Archive, as shown in formula (16).
[0101] P i =c / N i (16)
[0102] Where: c is a constant;
[0103] N i is the total number of individuals in the group to which the individual belongs.
[0104] When the multi-objective grey wolf algorithm is applied to the operation strategy optimization model (1) of the coal-fired unit coupled with the molten salt thermal storage system in this paper, the solution process is as follows: Figure 1 shown.
Claims
1. A method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method, characterized in that: Firstly, a system operation strategy optimization model is established based on a multi-objective evaluation method. The system heat rate and peak load capacity are set as dual-objective functions, and the unit constraints are considered. Then, a multi-objective grey wolf algorithm is used for optimization and solution. Finally, a set of solutions with the best comprehensive performance are output.
2. The method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method according to claim 1 is characterized in that: The optimization goal is to achieve the comprehensive optimum of the system heat rate and peak load capacity at the same time, that is, under the given boundary conditions of the heat storage process system operation, the system heat rate HC is minimized and the system peak load capacity ΔW is minimized. ch Maximize; in the optimization process, the peak load capacity ΔW ch Converted into unit output power, the goal is to achieve the lowest system output power and minimize the system heat rate. The objective function is shown in formula (1): Where: HC is the objective function of the system heat rate; W ch is the objective function of the system peak load capacity; T HT , T HTmin and T HTmax is the heat storage temperature parameter, the minimum and maximum values of the heat storage temperature, K; and are the thermal storage steam extraction flow parameters, the minimum and maximum thermal storage steam extraction flow rates, t / h; β, β min and β max is the split ratio parameter, the minimum and maximum split ratio; is an inequality constraint; is an equality constraint.
3. The method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method according to claim 2 is characterized in that: Introduce multidimensional preference analysis method to screen the optimal compromise solution and introduce the ideal point y * The concept of ideal point is the point composed of the optimal solutions of each single objective. This point is not located on the Pareto frontier and is an unattainable ideal point, as shown in formula (2): yes * =(y1 * ,y2 * ) (2) Where: y1 * is the optimal function value considering objective function 1; y2 * To consider the optimal function value of objective function 2; Among them, the multidimensional preference analysis method calculates the difference between each non-dominated solution on the Pareto front and the ideal point y * The Euclidean distance between them is used to select the point y′ with the shortest distance to the ideal point as the compromise optimal solution; the distance between the ideal point and any non-dominated solution in the solution set is the Euclidean distance, and the calculation formula is shown in formula (3): In the formula: y, y * are non-dominated solutions and ideal points respectively; y1,y1 * is the system heat rate function value of the non-dominated solution and the ideal optimal solution; y2,y2 * is the output power function value of the non-dominated solution and the ideal optimal solution.
4. The method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method according to claim 3 is characterized in that: Find the ideal point y in the non-dominated solution set * The point with the shortest Euclidean distance is taken as the compromise optimal solution y′, which is calculated using formula (4): The constraint violation degree when individual x violates the equality constraint is shown in formula (5), and the constraint violation degree when violating the inequality constraint is shown in formula (6): The total violation degree of individual x is shown in formula (7): By converting the unit constraints into penalty terms and adding them into the objective function, the constrained optimization problem shown in equation (1) is transformed into an unconstrained optimization problem shown in equation (8):
5. The method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method according to claim 4 is characterized in that: The multi-objective grey wolf algorithm is specifically: First, all individuals are sorted according to the fitness function. The optimal solution corresponds to the wolf α in the wolf pack. Two suboptimal solutions are selected as the wolf β and the wolf δ. The remaining feasible solutions are the wolf ω. The ω individual is optimized with the first three levels of individuals as the guide. Then, formulas (9) and (10) are used to imitate the hunting behavior of the wolf pack. Each ω individual uses an update mechanism to generate the next generation of solutions based on the positions of the three optimal solutions. Where: is the prey position vector; is the hunter position vector; is the coefficient vector; The hunter generates a new position vector based on the prey position and the random coefficient, as shown in formula (11): And formula (12) shows: Where: is a random parameter, taking a random number from [0,1] each time it is calculated; is the global convergence factor, which decreases from 2 to 0 during the iteration process; In the solution process, the optimal solution is estimated based on α, β, and δ individuals. The position of is shown in equations (13) and (14): Finally, the three updated positions are averaged using formula (15) as the result of this round of update; 6. The method for optimizing the operation strategy of a fire storage system based on a multi-objective evaluation method according to claim 5 is characterized in that: The multi-objective grey wolf algorithm has made two changes based on the grey wolf algorithm; The first is to introduce the concept of archive, which stores the non-dominated solutions generated in each iteration process; the second is to use the roulette strategy to select the top three gray wolves from the archive, and the probability of each individual being selected is P i It is inversely proportional to the number of individuals in the Archive, as shown in formula (16): P i =c / N i (16) Where: c is a constant; N i is the total number of individuals in the group to which the individual belongs.
