Optimization method of pumps and pipes in water system of nuclear power plant

Abstract

The invention discloses a method for optimizing pump pipes of a water system of a nuclear power plant. In the method, a model is constructed according to hydraulic characteristics and geometric connection relationships, input, output and optimization object are determined, the value range and step size of the optimization object are determined based on process conditions, and the operation state under multiple operating conditions during the whole life cycle is solved; based on the operation state, an optimization objective function and a design constraint function are constructed, and normalization and scale transformation processing are performed to obtain a fitness function as an evaluation index; an improved enhanced fast non-dominated sorting algorithm is used to divide the schemes into categories and levels, the non-dominated levels of schemes at different levels and the crowding distance of schemes at the same level are solved, and the advantages and disadvantages of the schemes under multi-dimensional objectives are evaluated; a multi-objective genetic algorithm and a differential evolution algorithm are respectively used for feasible domain schemes and infeasible domain schemes, and the parameters are iteratively optimized respectively, and the optimization results of the two algorithms are merged by an elite retention strategy to obtain the optimal scheme.

Description

Technical Field

[0001] The invention belongs to the technical field of nuclear power water systems, and in particular relates to a method for optimizing pump pipes of a nuclear power plant water system. Background Art

[0002] As one of the core technologies of intelligent operation and maintenance of nuclear power, the personalized customization technology of important nuclear power plant water systems has high automation, low labor intensity, low learning cost, good comprehensiveness, strong reliability, and high personalization. It is suitable for personalized, rapid and accurate design of large-scale equipment systems with many design objects, large parameter ranges, and high decision-making target dimensions. Due to the complexity of the SEC system design model, changes in parameters such as system flow, pipeline size, pump group model, and water loss parts configuration in actual projects may affect the safety and economy of the system. Therefore, the competitiveness of personalized customization technology for important nuclear power plant water systems needs to be enhanced urgently.

[0003] The parameters of nuclear power plant water pump pipes determine the hydraulic characteristics of the system, so the selection of pump pipe parameters is particularly important. The so-called pump pipe parameter design refers to the optimization of SEC pump models and pipe diameters and roughness based on system water intake capacity, equipment reliability and operation and maintenance costs, taking into account the changes in complex operating conditions throughout the life cycle. The traditional design method uses an expert decision-making system to optimize parameters through random trial and error, which is not only inefficient and has poor quality in generating solutions, but also has extremely high requirements on the professional quality of practitioners.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method for optimizing the pump pipe of a nuclear power plant water system, which can construct an adaptive threshold to divide the parameter space, and then match the appropriate genetic operator to meet different search capability requirements, avoiding the problem of "premature" or non-convergence in the optimization. At the same time, the threshold construction can introduce constraints under the framework of the traditional multi-objective optimization algorithm, making it suitable for multi-objective constrained optimization problems, avoiding the determination of pump pipe parameters by random trial and error, and thus effectively improving the comprehensive performance of the design scheme.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for optimizing a pump pipe of a water system of a nuclear power plant according to the present invention comprises:

[0008] The first step is to build a model of the seawater tunnel, trash rack, pre-pump water intake pipeline, pump room, SEC pump, heat exchanger and post-pump water intake pipeline that constitute the pipeline network according to the hydraulic characteristics and geometric connection relationship of the nuclear power plant water system. The model input includes two variables: pipeline network layout and offshore tide level. The decision-making objectives include five dimensions: cold source safety, pump overheating time, operation and maintenance cost, procurement cost, and civil construction cost. The design objects include six variables: seawater tunnel diameter, pre-pump water intake pipeline diameter, post-pump water intake pipeline diameter, pre-pump water intake pipeline diameter, pre-pump water intake pipeline roughness, and SEC pump model.

[0009] In the second step, the pipeline characteristic curve, pump family flow-head curve and pump family flow-efficiency curve under multiple working conditions during the whole life are solved based on the genetic algorithm to analyze the operating status of the pump pipe. The flow matrix Q in [0, 5000] is randomly generated using binary coding as the initial population; the average flow velocity v of the pipeline is calculated. i 、Head loss along the way h f and local head loss h j ,

[0010]

[0011] Where L and d represent the length and inner diameter of the pipeline, λ represents the head loss coefficient along the pipeline, and ζ represents the local head loss coefficient.

[0012] Construct fitness function Fit(x,Q)=|h f (x,Q)+h j (x, Q)-f Q-H (x, Q)|,

[0013] Where x represents the random solution generated based on the design object, f Q-H Indicates the pump flow-head characteristic curve,

[0014] The fitness is normalized and linearly scaled.

[0015]

[0016] Among them C mul represents the linear variation coefficient, F avg 、F max 、F min They represent the average fitness, maximum fitness and minimum fitness of the random scheme X corresponding to the flow Q respectively.

[0017] Based on the genetic algorithm, the population is continuously selected, crossed, and mutated until the termination condition is met, and the operating flow rate of the SEC under multiple working conditions during its entire life cycle is solved. The three-way bidding method is used to randomly select three flow values ​​from the flow matrix, and the excellent individuals with lower fitness are selected as the parents. The selected individuals are paired with each other and replaced and recombined according to the crossover probability to form offspring individuals. Some code positions of the generated offspring are mutated according to the mutation probability. If the fitness of the offspring after mutation is better, it will be retained, otherwise the individuals before mutation will continue to be preserved. Finally, the operating efficiency P of the SEC pump under multiple working conditions during the entire life cycle is obtained. L =f Q-p (x, Q, L), where f Q-p Indicates the pump flow-lift characteristic curve, L indicates five typical working conditions;

[0018] The third step is to construct an optimization objective function and a design constraint function based on the operating status and perform normalization and scale transformation processing to obtain a fitness function as an evaluation index, wherein procurement cost, civil engineering cost, and operating cost are used as optimization objective functions;

[0019]

[0020] Among them, f1, f2, f3 represent the operating cost, procurement cost and civil construction cost respectively, P1, P2, P3 represent the operating power of SEC pump under low tide, high frequency tide and high tide respectively, fre1, fre2, fre3 represent the tide frequency corresponding to low tide, high frequency tide and high tide respectively, 电 、F 泵 、F 管 、F 土 , T 电 They represent the unit price of power used in the power plant, the cost of pump power, the unit cost of pipeline, the unit price of earthwork and the life of the power plant respectively. L, d, h and ρ represent the length, diameter, thickness and material density of the pipeline respectively.

[0021] Pump family motor protection requirements and SEC water intake safety requirements as design constraint functions:

[0022] C1(x)=Relu(F 流量 -Q1), C2(x)=Relu(P4-F 功率 ),

[0023] Relu represents the activation function, which outputs the absolute value for negative numbers and 0 for positive numbers.

[0024] It indicates the flow rate of SEC pump at low tide and power at the highest tide. 流量 、F 功率Respectively represent the minimum flow rate and maximum power of a single pump. The constraint function is normalized according to the dimension and the constraint violation degree of the scheme is calculated:

[0025]

[0026] Normalize the objective function and perform linear fitness transformation according to the dimension.

[0027]

[0028] Among them C mult represents the linear variation coefficient, f i max 、f i avg 、f i min They represent the maximum objective function value, minimum objective function value and average objective function value of the solution set in the i-th dimension, respectively, and f′ i (x) represents the normalized objective function value, f″ i (x) represents the objective function value after linear scale change;

[0029] The fourth step is to use the improved enhanced fast non-dominated sorting algorithm to divide the solutions into categories and levels, and solve the non-dominated levels of solutions at different levels and the congestion distance of solutions at the same level, where:

[0030] According to the current optimization algebra, calculate the current constraint tolerance ε;

[0031]

[0032] Adopting the two-archive strategy, the population is divided into the feasible domain N0 and the infeasible domain N1;

[0033]

[0034] Arrange the solutions in descending order based on the objective function value of a specific dimension, and then determine the non-dominated levels of adjacent solutions one by one. The solution formula is as follows;

[0035]

[0036] Where N0 represents the feasible domain solution sequence in descending order, η represents its corresponding non-dominated level sequence, and f j (x) represents the corresponding objective function value of each dimension,

[0037] Arrange the solutions in descending order based on the constraint violation degree, and then determine the non-dominated level of adjacent solutions one by one. The solution formula is as follows;

[0038]

[0039] Where N1 represents the infeasible solution sequence in descending order, η′ represents its corresponding non-dominated level sequence, and C(x) represents the constraint violation degree of the solution.

[0040] According to the number of target dimensions, the schemes are sorted in ascending order. The crowding degree of individuals with selection advantages on the edge of the sort is set to 1. The crowding degree of individuals in the middle of the sort is calculated according to the formula. Finally, the crowding degree of individuals in different dimensions is summed up as the evaluation basis for the advantages and disadvantages of the same-layer schemes. The formula is as follows;

[0041]

[0042] In the fifth step, MOGA algorithm and DE algorithm are used for the feasible domain solution and the infeasible domain solution respectively. Through selection, crossover and mutation operations, the parameters are iteratively optimized respectively. The optimization results of the two algorithms are merged by using the elite retention strategy to obtain the optimal solution.

[0043] In the method for optimizing pump pipes in a nuclear power plant water system, in the fifth step, the feasible domain scheme uses the MODE algorithm to obtain the global optimal solution, wherein:

[0044] The three-way bidding method is used to randomly select three objects from the optimization scheme. The non-inferior levels of the individuals are compared first, and the individuals with the lowest level are selected first. If the levels are the same, the crowding degree is compared, and the individuals with a large crowding distance are selected first.

[0045] The selected individuals are paired in pairs, and replaced and recombined according to the crossover probability to form offspring individuals.

[0046] Some code bits that generate offspring will be mutated according to the mutation probability. If the non-dominated level and crowding degree of the offspring are better after mutation, they will be retained, otherwise the individuals before mutation will continue to be preserved.

[0047] In the method for optimizing the pump pipe of a nuclear power plant water system, a simulated binary crossover operator is used in the replacement and recombination to form offspring individuals, and a distribution coefficient η=1 is selected to generate a random number u j , generate new individuals according to the formula,

[0048]

[0049] In the method for optimizing the pump pipe of a nuclear power plant water system, when a mutation occurs, a polynomial mutation method is used to select a distribution index w=11 and generate a random number q j , operate according to the formula:

[0050]

[0051] In the method for optimizing the pump pipe of a water system of a nuclear power plant, in the fifth step, the DE algorithm is used in the infeasible region scheme to obtain a local optimal solution, wherein a differential strategy is used to realize individual variation, two different individuals are randomly selected from the population, their vector difference is calculated and processed by the differential scaling factor F, and finally vector synthesis is performed with the individual to be mutated to obtain an offspring individual.

[0052] H i (g+1)=X P1 (g)+F*(X P2 (g)-X P3 (g)),

[0053] Where X P1 (g), X P2 (g), X P3 (g) represents the individuals to be optimized that are randomly selected during the g-th generation optimization process. The selected individuals are paired with each other and replaced and recombined according to the crossover probability to form offspring individuals.

[0054] A greedy strategy is used to form the next generation population. After each crossover and mutation, the non-dominated level and crowding degree of the parent and offspring individuals are compared, and the best individuals are retained to enter the next generation population.

[0055]

[0056] Where V i (g), X i (g), respectively represent the offspring individuals and parent individuals randomly selected in the i-th selection, X i (g+1) represents the next generation of individuals generated, η(x) represents the non-dominated level of individuals,

[0057] Introduce the elite retention strategy: construct a recorder T to store the optimal solution in the optimization process. The optimal solution is a feasible solution with a non-dominated level of 0 and an infeasible solution with a non-dominated level of k. After each parameter optimization is completed, the optimization results and individuals in T are re-performed with the improved enhanced fast non-dominated sorting in step 4, the better solutions are retained to form the offspring population, and the recorder T is updated. Repeat the operation to gradually optimize the solution parameters.

[0058] In the above technical scheme, the present invention provides a nuclear power plant water system pump pipe optimization method, which has the following beneficial effects: the present invention constructs a SEC multi-objective optimization model, including a simplified analysis of system layout, operating conditions, and decision-making objectives; predicts the operating state of the SEC pump under complex operating conditions throughout the life cycle, solves the target fitness and constraint violation of the solution, and performs normalization and scale change processing; proposes an improved enhanced fast non-dominated sorting method, and evaluates the advantages and disadvantages of the solution under multi-dimensional objectives based on adaptive safety redundancy; finally, according to the proposed heuristic hybrid intelligent algorithm, an intelligent algorithm with matching search capabilities is selected according to the characteristics of the solution for parameter optimization. Aiming at the problem that the multi-objective optimization algorithm is not suitable for constrained optimization, the present invention proposes a constrained optimization algorithm framework based on the improved enhanced fast non-dominated sorting method, which can solve the constrained optimization problem without making any changes to the traditional method, and has strong applicability; Aiming at the single search capability of the traditional multi-objective optimization method, which is prone to "premature" or non-convergence defects, a heuristic hybrid intelligent method is proposed, which effectively improves the efficiency and convergence of the algorithm. At the same time, the present invention provides a fast and intelligent method for the multi-objective optimization design of SEC, and provides a target orientation for the subsequent SEC cold source safety and shore siltation analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0060] Figure 1 It is a schematic flow diagram of the pump pipe optimization method of the nuclear power plant water system in the present invention;

[0061] Figure 2 This is a comparison chart of the optimization results of the proposed method based on the international standard test function OSY and the MOGA and DE methods;

[0062] Figure 3 It is a performance index comparison chart of the proposed method and MOGA and DE methods based on the international standard test function OSY;

[0063] Figure 4 A schematic diagram of solving pipeline characteristics of an embodiment of a method for optimizing pump pipes in a water system of a nuclear power plant in the present invention;

[0064] Figure 5 A schematic diagram of an error variation curve for solving pump power based on pipeline characteristics in one embodiment of a nuclear power plant water system pump pipe optimization method of the present invention;

[0065] Figure 6It is a schematic diagram of HV variation curve of an embodiment of the method for optimizing pump pipes of a water system of a nuclear power plant in the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Therefore, the following Figures 1 to 6 The detailed description of the embodiments of the present invention provided in the present invention is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention.

[0068] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0069] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0070] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0071] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be directly connected or indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0072] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply indicates that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply indicates that the first feature is lower in level than the second feature.

[0073] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings. Figures 1 to 6 As shown, a method for optimizing pump pipes in a water system of a nuclear power plant includes:

[0074] Step 1: According to the hydraulic characteristics and geometric connection relationship of the system, a simplified mathematical model is constructed to determine the design input, output and optimization object, and the value range and step size of the optimization object are determined based on the process conditions.

[0075] Step 2: Based on the Hydraulic Calculation Manual and genetic algorithms, solve the pipeline characteristic curve, pump family flow-head curve and pump family flow-efficiency curve under multiple working conditions during the whole life cycle, and accurately analyze the operating status of the pump pipe;

[0076] Step 3: Based on the operating status, construct the optimization objective function and design constraint function, and perform normalization and scale transformation to obtain the fitness function as an evaluation index;

[0077] Step 4: Use the improved enhanced fast non-dominated sorting algorithm to divide the solutions into categories and levels, solve the non-dominated levels of solutions at different levels and the congestion distance of solutions at the same level, and evaluate the advantages and disadvantages of solutions under multi-dimensional objectives;

[0078] Step 5: Adopt heuristic hybrid intelligent algorithm, use multi-objective genetic algorithm and differential evolution for feasible domain solution and infeasible domain solution respectively, and iteratively optimize the parameters through selection, crossover and mutation operations. Finally, use elite retention strategy to merge the optimization results of the two algorithms and get the optimal solution.

[0079] The simplified mathematical model and related parameter steps described in step 1 are as follows:

[0080] The mathematical model includes the simplification of SEC layout and the simplification of the whole life cycle operating conditions. According to the hydraulic characteristics and geometric connection relationship of the system, SEC is simplified into seven key equipment: seawater tunnel, trash rack, water intake pipeline before pump, pump room, SEC pump, heat exchanger, and each object is connected through local water loss parts to form a pipeline network; according to expert experience, the complex variable operating conditions during the whole life cycle are simplified into five typical operating conditions, including high-frequency tide level, high tide level, low tide level, lowest tide level, and highest tide level.

[0081] The pipeline network layout and offshore tide records were selected as design inputs, and six variables in the pump pipe process parameters were selected as design objects, including the seawater tunnel diameter, the water intake pipe diameter before the pump, the water intake pipe roughness before the pump, the water intake pipe diameter after the pump, the water intake pipe roughness after the pump, and the SEC pump type. Five variables in the SEC evaluation index were selected as design outputs, including civil construction cost, operating cost, procurement cost, SEC pump maximum power, and SEC pump minimum efficiency.

[0082] The operation status steps based on the full life cycle multi-operating condition prediction scheme described in step 2 are as follows:

[0083] (1) Using binary coding, setting the number of significant digits to two decimal places, and randomly generating a traffic matrix Q in [0, 5000] as the initial population;

[0084] (2) Solve the average flow velocity v of the pipe flow section i . According to the Darcy-Weisbach formula, the head loss h along the pipeline is solved f , according to the formula in the "Hydraulic Calculation Manual", solve the local head loss of local water loss parts.

[0085]

[0086] Where L is the length of the pipeline, d is the inner diameter of the pipeline, λ is the head loss coefficient along the pipeline, and ζ is the local head loss coefficient, which can be determined by referring to the Hydraulic Calculation Manual.

[0087] (3) Construct the fitness function Fit(x, Q):

[0088] Fit(x,Q)=|h f (x,Q)+h j(x, Q)-f Q-H (x, Q)|

[0089] x represents the random solution generated based on the design object, f Q-H Indicates the pump flow-head characteristic curve.

[0090] (4) The fitness is normalized and linearly scaled to reduce the differences between individuals and prevent the algorithm from falling into the local optimal solution too early.

[0091]

[0092] Among them C mult Indicates the linear variation coefficient, F avg 、F max 、F min They represent the average fitness, maximum fitness and minimum fitness of scheme X corresponding to flow Q respectively.

[0093] (5) Based on the genetic algorithm, the population is continuously selected, crossed, and mutated until the termination condition is met, and the operating flow rate of the SEC under multiple operating conditions throughout its life cycle is accurately solved.

[0094] ① Selection: According to the survival of the fittest mechanism, a ternary bidding method is used to randomly select three flow values ​​from the flow matrix, and select excellent individuals with lower fitness as parents.

[0095] ② Crossover: Pair the selected individuals in pairs according to the crossover probability P c =0.6 to replace and recombine it to form offspring individuals. This patent adopts the method of single-point crossover operator, that is, randomly generates one crossover point to replace genes with each other.

[0096]

[0097] Where U represents the uniform distribution of values ​​in [0, 1], and Q[k] represents its kth binary code bit.

[0098] ③ Mutation: Generate some code bits of the offspring according to the mutation probability P m =0.2 for mutation. If the offspring has better fitness after mutation, it will be retained. Otherwise, the individual before mutation will continue to be preserved. This patent adopts the method of single-point position flip mutation, that is, randomly generating a mutation point to flip the value of the gene. The mutation method is as follows:

[0099]

[0100] Where U represents the uniform distribution of values ​​in [0, 1], and Q[n] represents its nth binary code position.

[0101] (6) Based on the flow-efficiency characteristic curve of the pump family, the operating efficiency of the SEC pump under multiple operating conditions throughout its life cycle is solved.

[0102] P L =f Q-p (x, Q, L)

[0103] where f Q-P represents the pump flow-lift characteristic curve, L represents the five typical working conditions in 2,

[0104] The steps of predictive analysis based on the running status of the scheme described in step 3, constructing the objective and constraint functions, and performing normalization and linear scale change are as follows:

[0105] (1) Select three variables in the scheme evaluation index: procurement cost, construction cost, and operating cost as the optimization objective function;

[0106]

[0107] Among them, f1, f2, f3 represent the operating cost, procurement cost and civil construction cost respectively, P1, P2, P3 represent the operating power of SEC pump under low tide, high frequency tide and high tide respectively, fre1, fre2, fre3 represent the tide frequency corresponding to low tide, high frequency tide and high tide respectively, 电 、F 泵 、F 管 、F 土 , T 电 They represent the unit price of plant electricity, pump power cost, unit pipeline cost, unit earthwork price and power plant life respectively. L, d, h, ρ represent the length, diameter, thickness and material density of the pipeline respectively.

[0108] (2) Select two variables in the scheme evaluation index: pump motor protection requirements and SEC water intake safety requirements as design constraint functions:

[0109] C1(x)=Relu(F 流量 -Q1), C2(x)=Relu(P4-F 功率 )

[0110] Relu is an activation function, which outputs the absolute value for negative numbers and 0 for positive numbers. Q1 and P4 represent the flow rate at low tide and the power at the highest tide of the SEC pump, respectively. 流量 、F 功率 Respectively represent the minimum flow rate and maximum power of a single pump.

[0111] (3) Normalize the constraint function according to the dimension to avoid a single indicator dominating the scheme evaluation and calculate the constraint violation degree of the scheme;

[0112]

[0113] (4) Normalize the objective function and perform linear fitness transformation according to the dimension to reduce the differences between individuals and enhance the global search capability of the algorithm:

[0114]

[0115] Among them C mult represents the linear variation coefficient, f i max 、f i avg 、f i min They represent the maximum objective function value, minimum objective function value and average objective function value of the solution set in the i-th dimension, respectively, and f′ i (x) represents the normalized objective function value, f″ i (x) represents the objective function value after linear scale change.

[0116] The steps for evaluating the merits of the scheme based on the improved enhanced fast non-dominated sorting algorithm described in step 4 are as follows:

[0117] This patent combines the classic ε strategy in constrained optimization to improve the traditional enhanced fast non-dominated sorting algorithm to make it suitable for multi-objective constrained optimization problems. The specific method is as follows

[0118] (1) Calculate the current constraint tolerance ε based on the current optimization algebra;

[0119]

[0120] (2) Adopting the two-archive strategy, the population is divided into the feasible domain N0 and the infeasible domain N1;

[0121]

[0122] (3) The fast non-dominated sorting method is used to judge the superiority of feasible solutions. First, the solutions are sorted in descending order based on the objective function value of a specific dimension, and then the non-dominated levels of adjacent solutions are determined one by one. The solution formula is as follows;

[0123]

[0124] Where N0 represents the feasible domain solution sequence in descending order, η represents its corresponding non-dominated level sequence, and f j (x) represents the corresponding objective function value of each dimension.

[0125] (4) The constraint dominance ranking method is used to judge the superiority and inferiority of the infeasible region scheme. First, the schemes are sorted in descending order based on the constraint violation degree, and then the non-dominated level of adjacent schemes is judged one by one. The solution formula is as follows;

[0126]

[0127] Where N1 represents the infeasible solution sequence in descending order, η′ represents its corresponding non-dominated level sequence, and C(x) represents the constraint violation degree of the solution.

[0128] (5) Analyze the advantages and disadvantages of the same non-dominated level schemes based on the crowding distance. According to the number of target dimensions, the schemes are sorted in ascending order. The crowding degree of individuals with selection advantages on the edge of the sort is set to 1. The crowding degree of individuals in the middle of the sort is calculated according to the formula. Finally, the crowding degree of individuals in different dimensions is summed up as the evaluation basis for the advantages and disadvantages of the same-level schemes. The formula is as follows;

[0129]

[0130] The steps of using the heuristic hybrid intelligent algorithm to optimize parameters of different types of solutions in step 5 are as follows:

[0131] (1) The feasible domain solution uses the MODE algorithm to fully solve the global optimal solution with strong search capabilities

[0132] ① Selection: According to the survival of the fittest mechanism, three objects are randomly selected from the optimization schemes using the ternary bidding method. The non-inferior levels of the individuals are compared first, and the individuals with the lowest level are selected first. If the levels are the same, the crowding degree is compared, and the individuals with a large crowding distance are selected first.

[0133] ② Crossover: Pair the selected individuals in pairs according to the crossover probability P c =0.8 to replace and reorganize the offspring individuals. This patent uses a simulated binary crossover operator. Select the distribution coefficient η = 1 and generate a random number u j , generate new individuals according to the formula

[0134]

[0135] ③ Mutation: Generate some code bits of the offspring according to the mutation probability P m = 0.2 for mutation. If the non-dominated level and crowding degree of the offspring after mutation are better, they will be retained. Otherwise, the individuals before mutation will continue to be preserved. This patent adopts the polynomial mutation method, selects the distribution index w = 11, and generates a random number q j .

[0136] Operate according to the formula:

[0137]

[0138] (2) The infeasible region solution uses the DE algorithm to fully solve the local optimal solution with strong convergence ability.

[0139] ① Mutation: Use the differential strategy to achieve individual mutation. Randomly select two different individuals from the population, calculate their vector difference and process it by the differential scaling factor F, and finally synthesize the vector with the individual to be mutated to obtain the offspring individual, that is,

[0140] H i (g+1)=X P1 (g)+F*(X P2 (g)-X P3 (g))

[0141] Where X P1 (g), X P2 (g), X P3 (g) represents the individuals to be optimized randomly selected during the g-th generation optimization process, and F=0.8.

[0142] ② Crossover: Pair the selected individuals in pairs, and replace and recombine them according to the crossover probability CR = 0.4 to form offspring individuals. This patent adopts the single-point crossover operator, that is, randomly generate a crossover point to replace genes.

[0143]

[0144] Where j represents the jth code point of the floating point encoding of the scheme parameter.

[0145] ③ Selection: Use the greedy strategy to form the next generation population. After each crossover and mutation, compare the non-dominated level and crowding degree of the parent and offspring individuals, and retain the best individuals to enter the next generation population, that is,

[0146]

[0147] Where V i (g), X i (g), respectively represent the offspring individuals and parent individuals randomly selected in the i-th selection, X i (g+1) represents the next generation individual generated, and η(x) represents the non-dominated level of the individual.

[0148] (3) Introduce the elite retention strategy: Construct a recorder T to store the optimal solution in the optimization process, that is, the feasible solution with a non-dominated level of 0 and the infeasible solution with a non-dominated level of k. After each parameter optimization is completed, the optimization results and individuals in T are re-performed with the improved enhanced fast non-dominated sorting in step 4, retain the better solutions to form the offspring population, and update the recorder T. In this way, while ensuring that the number of populations in each generation remains unchanged, it is ensured that the excellent individuals of the previous generation will not be destroyed by operations such as crossover mutation, which is conducive to the evolution of excellent individual genes.

[0149] Repeat the above selection, crossover, mutation, and fusion operations to gradually optimize the scheme parameters. After reaching the termination condition, a design scheme with better economy and safety is obtained.

[0150] Finally, it should be noted that the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making any creative work are within the scope of protection of the present application.

[0151] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing pump pipes in a nuclear power plant water system, characterized in that: It includes the following steps: The first step is to build a design model consisting of a seawater tunnel, trash rack, pre-pump water intake pipeline, pump room, SEC pump, heat exchanger and post-pump water intake pipeline according to the hydraulic characteristics and geometric connection relationship of the nuclear power plant water system. The model input includes two variables: pipeline network layout and offshore tide level. The decision-making objectives include five dimensions: cold source safety, pump overheating time, operation and maintenance cost, procurement cost, and civil construction cost. The design objects include six variables: seawater tunnel diameter, pre-pump water intake pipeline diameter, post-pump water intake pipeline diameter, pre-pump water intake pipeline diameter, pre-pump water intake pipeline roughness, and SEC pump model. In the second step, the pipeline characteristic curve, pump family flow-head curve and pump family flow-efficiency curve under multiple working conditions during the whole life are solved based on the genetic algorithm to analyze the operating status of the pump pipe. The flow matrix Q in [0,5000] is randomly generated by binary coding as the initial population; the average flow velocity v of the pipeline is calculated. i 、Head loss along the way h f and local head loss h j , Where L and d represent the length and inner diameter of the pipeline, λ represents the head loss coefficient along the pipeline, and ζ represents the local head loss coefficient; Construct fitness function Fit(x,Q)=|h f (x,Q)+h j (x,Q)-f Q-H (x,Q)|, Where x represents the random solution generated based on the design object, f Q-H Indicates the pump flow-head characteristic curve, The fitness is normalized and linearly scaled. Among them C mul Indicates the linear variation coefficient, F avg 、F max 、F min They represent the average fitness, maximum fitness and minimum fitness of the random scheme X corresponding to the flow Q, Based on the genetic algorithm, the population is continuously selected, crossed, and mutated until the termination condition is met. The operating flow of the nuclear power plant water system under multiple conditions during the whole life cycle is solved. The three-way bidding method is used to randomly select three flow values ​​from the flow matrix, and the excellent individuals with lower fitness are selected as the parent generation. The selected individuals are paired with each other, and they are replaced and recombined according to the crossover probability to form offspring individuals. Some code bits of the generated offspring are mutated according to the mutation probability. If the fitness of the offspring is better after the mutation, it will be retained. Otherwise, the individuals before the mutation will continue to be preserved. Finally, the SEC pump operating efficiency P under multiple conditions during the whole life cycle is obtained. L =f Q-p (x,Q,L), where f Q-P Indicates the pump flow-lift characteristic curve, L indicates five working conditions; The third step is to construct an optimization objective function and a design constraint function based on the operating status and perform normalization and scale transformation processing to obtain a fitness function as an evaluation index, wherein procurement cost, civil engineering cost, and operating cost are used as optimization objective functions; Among them, f1, f2, f3 represent the operating cost, procurement cost and civil construction cost respectively, P1, P2, P3 represent the operating power of SEC pump under low tide, high frequency tide and high tide respectively, fre1, fre2, fre3 represent the tide frequency corresponding to low tide, high frequency tide and high tide respectively, 电 、F 泵 、F 管 、F 土 、T 电 They represent the unit price of power used in the power plant, the cost of pump power, the unit cost of pipeline, the unit price of earthwork and the life of the power plant respectively. L, d, h and ρ represent the length, diameter, thickness and material density of the pipeline respectively. Pump family motor protection requirements and SEC water intake safety requirements as design constraint functions: C1(x)=Relu(F 流量 -Q1),C2(x)=Relu(P4-F 功率 ), Relu represents the activation function, which outputs the absolute value for negative numbers and 0 for positive numbers. Q1 and P4 represent the flow rate of the SEC pump at low tide and the power at the highest tide respectively. 流量 、F 功率 They represent the minimum flow rate and maximum power of a single pump respectively, normalize the constraint function according to the dimension, and calculate the constraint violation degree of the scheme; Normalize the objective function and perform linear fitness transformation according to the dimension. Among them C mult represents the linear variation coefficient, f i max 、f i avg 、f i min Respectively represent the maximum objective function value, minimum objective function value and average objective function value of the solution set in the i-th dimension, f' i (x) represents the normalized objective function value, f” i (x) represents the objective function value after linear scale change; The fourth step is to use the improved enhanced fast non-dominated sorting algorithm to divide the solutions into categories and levels, and solve the non-dominated levels of solutions at different levels and the congestion distance of solutions at the same level, where: According to the current optimization algebra, calculate the current constraint tolerance ε; Adopting the two-archive strategy, the population is divided into the feasible domain N0 and the infeasible domain N1; Arrange the feasible domain solutions in descending order based on the objective function value of a specific dimension, and then determine the non-dominated level of adjacent solutions one by one. The solution formula is as follows; Where N0 represents the feasible domain solution sequence in descending order, η represents its corresponding non-dominated level sequence, and f j (x) represents the corresponding objective function value of each dimension, Arrange the infeasible solutions in descending order based on the constraint violation degree, and then judge the non-dominated level of adjacent solutions one by one. The solution formula is as follows; Where N1 represents the infeasible solution sequence in descending order, η' represents its corresponding non-dominated level sequence, and C(x) represents the constraint violation degree of the solution. According to the number of target dimensions, the schemes are sorted in ascending order. The crowding degree of individuals with selection advantages on the edge of the sort is set to 1. The crowding degree of individuals in the middle of the sort is calculated according to the formula. Finally, the crowding degree of individuals in different dimensions is summed up as the evaluation basis for the advantages and disadvantages of the same-layer schemes. The formula is as follows; In the fifth step, MOGA algorithm and DE algorithm are used for the feasible domain solution and the infeasible domain solution respectively. Through selection, crossover and mutation operations, the parameters are iteratively optimized respectively. The optimization results of the two algorithms are merged by using the elite retention strategy to obtain the optimal solution.

2. A method for optimizing pump pipes in a nuclear power plant water system according to claim 1, characterized in that: Preferably, in the fifth step, the feasible domain scheme adopts the MODE algorithm to obtain the global optimal solution, wherein: The three-way bidding method is used to randomly select three objects from the optimization scheme. The non-inferior levels of the individuals are compared first, and the individuals with the lowest level are selected first. If the levels are the same, the crowding degree is compared, and the individuals with a large crowding distance are selected first. The selected individuals are paired in pairs, and replaced and recombined according to the crossover probability to form offspring individuals. Some code bits that generate offspring will be mutated according to the mutation probability. If the non-dominated level and crowding degree of the offspring are better after mutation, they will be retained, otherwise the individuals before mutation will continue to be preserved.

3. A method for optimizing pump pipes in a nuclear power plant water system according to claim 2, characterized in that: The simulated binary crossover operator is used in the replacement recombination to form the offspring individuals, the distribution coefficient η = 1 is selected, and the random number u is generated j , generate new individuals according to the formula, 4. A method for optimizing pump pipes in a nuclear power plant water system according to claim 3, characterized in that: When mutation occurs, the polynomial mutation method is used, the distribution index w=11 is selected, and the random number q is generated j , operate according to the formula: x 1j =x 1j (t)+Δj, 5. A method for optimizing pump pipes in a nuclear power plant water system according to claim 1, characterized in that: In the fifth step, the DE algorithm is used to obtain the local optimal solution for the infeasible region solution. The differential strategy is used to achieve individual mutation. Two different individuals are randomly selected from the population, their vector difference is calculated and processed by the differential scaling factor F, and finally the vector is synthesized with the individual to be mutated to obtain the offspring individual. H i (g+1)=X P1 (g)+F*(X P2 (g)-X P3 (g)), Where X P1 (g), X P2 (g), X P3 (g) represents the individuals to be optimized that are randomly selected during the g-th generation optimization process. The selected individuals are paired with each other and replaced and recombined according to the crossover probability to form offspring individuals. A greedy strategy is used to form the next generation population. After each crossover and mutation, the non-dominated level and crowding degree of the parent and offspring individuals are compared, and the best individuals are retained to enter the next generation population. Where V i (g), X i (g) represent the randomly selected offspring individuals and parent individuals in the i-th selection, respectively, and X i (g+1) represents the next generation of individuals generated, η(x) represents the non-dominated level of individuals, Introduce the elite retention strategy: construct a recorder T to store the optimal solution in the optimization process. The optimal solution is a feasible solution with a non-dominated level of 0 and an infeasible solution with a non-dominated level of k. After each parameter optimization is completed, the optimization results and individuals in T are re-performed with the improved enhanced fast non-dominated sorting in step 4, the better solutions are retained to form the offspring population, and the recorder T is updated. Repeat the operation to gradually optimize the solution parameters.

Images