Water, wind and light optimal scheduling method and system based on two-stage double-population evolutionary algorithm

Abstract

The application belongs to the field of water, wind and light multi-ability complementary optimal scheduling, and particularly discloses a water, wind and light optimal scheduling method and system based on a two-stage double-population evolution algorithm, which comprises the following steps: constructing a multi-objective scheduling model by taking reservoir water level as a decision variable and taking maximum power generation and minimum residual load mean square deviation as objective functions; initializing two populations and solving the water, wind and light multi-objective scheduling model; the two populations are in a fixed division stage at the beginning and are respectively updated by operators A and B; when the switching condition is met, the two populations are switched to an adaptive cooperation stage for iterative updating, at this time, the selection probability of operators A and B is determined according to the population performance, and the operators are selected based on the selection probability for iterative updating; the switching condition is that the population diversity is less than a threshold value or the evaluation number reaches a threshold value after each iterative updating of the population, and the stage is switched. The application can solve the contradiction between convergence and diversity in water, wind and light multi-objective optimization and realize precise optimization.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of water, wind and light multi-energy complementary optimal scheduling, and more particularly to a water, wind and light optimal scheduling method and system based on a two-stage double-population evolutionary algorithm. BACKGROUND

[0002] The core of multi-objective optimal scheduling of a water, wind and light multi-energy complementary system is to solve the competitive trade-off problem between power generation benefits on the power supply side and peak shaving demand on the power grid side. It is necessary to both improve the total power generation of hydropower stations, wind farms and photovoltaic power stations to ensure economic efficiency and reduce power grid residual load fluctuations to improve power supply stability. However, the improvement of power generation is often accompanied by an increase in residual load fluctuations, and the power generation and residual load fluctuation objectives are coupled and conflict. This problem is a typical dynamic nonlinear multi-objective optimization problem. However, traditional methods have obvious limitations in solving the water, wind and light multi-energy complementary optimal scheduling problem.

[0003] (1) Existing water, wind and light optimal scheduling technologies mainly use single population evolutionary algorithms (such as NSGA-II, MOPSO and SPEA2). All individuals follow a unified evolutionary rule, making it difficult to balance global exploration and local exploitation needs, and prone to premature loss of population diversity or slow convergence, which cannot adapt to the dynamic needs of wide-area exploration of high-quality solution space in the early stage of water, wind and light scheduling and precise optimization of local areas in the later stage. Different operator update mechanisms often have different focuses. If the focus is on global exploration, it is easy to lead to slow convergence in the later stage and make it difficult to obtain a high-precision scheduling scheme. If the focus is on local exploitation, it is easy to cause premature loss of population diversity and fall into local optimization.

[0004] (2) Although some technologies use a double population strategy or a multi-operator method (such as a multi-objective trading strategy and system for water, wind and light power generation based on an improved hybrid algorithm), the double population only uses different algorithms (such as MOPSO and NSGAII) to independently optimize and update the population, and only a part of the population is archived and migrated in the later stage. There is no exponential decay scoring and dynamic selection probability distribution logic for the real-time success rate of the operator, making it difficult to adapt to the engineering characteristics and multi-objective collaborative optimization needs of water, wind and light multi-objective scheduling. Traditional multi-operator algorithms use fixed probabilities or random methods to select operators (such as GA and DE), which cannot adjust according to the optimization process and real-time performance of the operator, leading to mismatched search resources and low efficiency. For example, if global exploration is needed, a large number of local search operators are still used, or if local optimization is needed, global operators are still relied on.

[0005] (3) Most collaborative evolutionary algorithms only focus on horizontal information exchange between populations, ignoring the vertical guiding role of historical elite solutions in the external archive. Historical elite solutions carry key high-quality genes, leading to insufficient information utilization, easy loss of high-quality genes and reduced convergence stability.

[0006] (4) The prior art all adopts a single strategy. If the selection strategy of NSGA-II is used throughout, selection pressure is insufficient under the high-dimensional objective space or complex constraints of water, wind and light scheduling, resulting in lack of solution set diversity, and it is difficult to provide rich power generation and peak shaving trade-off schemes. If the selection strategy of SPEA2 is used throughout, the calculation of the dominance strength and the distance between individuals in the objective space needs to be calculated additionally, the calculation overhead is significantly increased, and the real-time requirement of engineering scheduling cannot be met. Both of them are difficult to balance, which becomes the key bottleneck restricting the algorithm to adapt to the water, wind and light multi-objective scheduling scene.

[0007] In summary, there is an urgent need for an optimization method that can dynamically balance exploration and exploitation, adaptively match operator resources, efficiently utilize elite information, and balance efficiency and quality. SUMMARY

[0008] In view of the above defects or improvement needs of the prior art, the present application provides a water, wind and light optimal scheduling method and system based on a two-stage double-population evolutionary algorithm, which aims to solve the contradiction between convergence and diversity in water, wind and light multi-objective optimization and achieve accurate optimization.

[0009] To achieve the above-mentioned purpose, according to one aspect of the present application, a water, wind and light optimal scheduling method based on a two-stage double-population evolutionary algorithm is proposed, comprising the following steps:

[0010] A water, wind and light multi-objective scheduling model is constructed with reservoir water level as the decision variable and maximum power generation and minimum residual load variance of the water, wind and light multi-energy complementary system as the objective function;

[0011] Two populations are initialized, and the water, wind and light multi-objective scheduling model is solved. The two populations are initially in a fixed division stage, and when the stage switching condition is met, they are switched to an adaptive cooperation stage for iterative updating, and finally a solution set is obtained, thereby realizing water, light and wind optimal scheduling; wherein:

[0012] Fixed division stage: two populations are iteratively updated by operators A and B respectively;

[0013] Adaptive cooperation stage: the selection probability of operators A and B is determined according to the real-time performance of the population, and two populations select operators A or B based on the selection probability for iterative updating;

[0014] The stage switching condition is that in the fixed division stage, the population diversity is evaluated after each iterative update, and when the population diversity is less than the diversity threshold or the number of iterative evaluations reaches the evaluation number threshold, the stage is switched.

[0015] As a further optimization, the selection probability of operators A and B is determined according to the real-time performance of the population, including:

[0016] For the operators A and B, the proportion of the offspring generated by the operator to be reserved by the next generation population is taken as the success rate of the operator;

[0017] Based on the success rate of the operator, an attenuation mechanism is used to update the score of the operator:

[0018]

[0019] wherein, t is the current iteration number, , are respectively the score and the success rate of the i-th generation operator, t is the operator type, is the attenuation coefficient; According to the proportion of the score of the operator in the total score of the two operators, the selection probability of the operator is obtained.

[0020] According to the proportion of the score of the operator in the total score of the two operators, the selection probability of the operator is obtained.

[0021] As a further preferred, the method for determining the population diversity is: the diversity of the two populations is calculated respectively by the following formula, and then the mean of the diversity of the two populations is taken as the final population diversity:

[0022]

[0023] wherein, is the diversity of a single population, and N is the population size, , are respectively the objective function values of the i-th and j-th individuals in the population, is the Euclidean distance operator.

[0024] As a further preferred, after each population is updated in the adaptive cooperation stage, the top 10% of elite solutions are selected from the other population and the top 10% of elite solutions are selected from the external archive, and then the union of the two types of elite solutions is migrated to the current population and participates in the next environmental selection.

[0025] As a further preferred, the screening method of the elite solution is: first, the individuals in the migration source are arranged in ascending order according to the non-dominated level, and then the individuals in the migration source are arranged in descending order according to the crowding distance, and then the top 10% of individuals in the sorted migration source are selected as the elite solution; the migration source is the other population or the external archive.

[0026] As a further preferred, the updating method of the external archive is: every K generations are selected from the population to update the external archive according to the Pareto strength and the Euclidean distance, and the rest of every generation are selected from the population to update the external archive according to the non-dominated level and the crowding distance.

[0027] ​As a further preferred, the individuals are screened from the population according to non-dominated rank and crowding distance, specifically:

[0028] The population is non-dominantly ranked to obtain several non-dominated layers ranked from high to low, the individuals in the non-dominated layers are sequentially put into the external archive according to the ranking, the external archive stops when reaching the archive size, and for the non-dominated layer that just reaches the archive size, the crowding distance of all individuals in the layer is calculated, and part of the individuals are selected based on the crowding distance and put into the external archive to make the individuals in the external archive reach the archive size.

[0029] As a further preferred, the individuals are screened from the population according to Pareto strength and Euclidean distance truncation, specifically:

[0030] The individuals in the population with a Pareto strength value less than 1 are screened out;

[0031] If the number of screened individuals does not exceed the archive size, all the screened individuals are directly stored in the external archive;

[0032] If the number of screened individuals exceeds the archive size, the Euclidean distance between the screened individuals is calculated first, and a distance matrix is constructed, the crowding degree of each individual is calculated based on the distance matrix, and the most crowded individual is deleted, then the above process is repeated until the number of remaining individuals matches the archive size, and the remaining individuals are stored in the external archive.

[0033] As a further preferred, the operators A and B are genetic operators and difference operators respectively.

[0034] According to another aspect of the present application, a water, wind and light optimization scheduling system based on a two-stage double population evolutionary algorithm is provided, comprising a processor, the processor being used to execute the water, wind and light optimization scheduling method based on the two-stage double population evolutionary algorithm.

[0035] Overall, compared with the prior art, the above technical solutions conceived by the present application mainly have the following technical advantages:

[0036] 1. Unlike the traditional multi-objective algorithm using a single population, the present application breaks through the traditional single population evolution mode, designs an evolution system composed of two parallel populations, balances the global exploration and local exploitation ability through the dynamic switching from fixed division exploration stage to adaptive collaborative development stage, and solves the contradiction between convergence and diversity in multi-objective optimization.

[0037] 2、Traditional multi-operator algorithm mostly adopts fixed probability or random way to select operator, which cannot dynamically adjust according to optimization process and real-time performance of the operator, and the search efficiency is low. The application proposes a dynamic operator selection strategy based on the success rate under the screening of different historical environments, dynamically adjusts the selection probability of the two operators, the higher the proportion of the operator to produce high-quality offspring and be retained, the higher the score, and the greater the probability of being selected subsequently, so that the algorithm autonomously learns the optimal operator in the current optimization stage, and the evolution efficiency is improved.

[0038] 3、Most of the co-evolution algorithms have limitations, only focusing on information exchange between populations, but ignoring the guiding role of historical elite solutions, resulting in insufficient information utilization. To break through this limitation, the application designs a two-way, multi-source, multi-level elite migration mechanism, realizing the organic combination of horizontal migration between populations and vertical migration of external archive; the mechanism not only promotes horizontal communication between populations and accelerates the spread of excellent modes and genes among populations, but also realizes the vertical inheritance of historical elites, effectively avoiding the problem of extinction of excellent genes due to the population falling into local optimum. Ultimately, not only the convergence speed of the algorithm is significantly improved, but also the population diversity is enriched, and the robustness and global search ability of the algorithm are enhanced.

[0039] 4、In order to solve the problem that the single environment selection strategy cannot comprehensively consider the calculation efficiency and solution set quality, the application designs a mixed selection system with different scenes and differentiations, which adapts the core advantages of the two strategies to different scenes. For the update of the external archive of each generation, the fast non-dominated sorting and crowding distance strategy with high calculation efficiency is adopted to preferentially ensure the iteration speed of the algorithm and meet the timeliness demand of water, light and scenery scheduling; for the maintenance of the external archive, the Pareto intensity calculation and minimum distance truncation strategy are triggered regularly, which utilizes the dual advantages of diversity and convergence to accurately maintain the uniformity and coverage of historical elite solutions, so as to ensure that the high-quality solutions with wide coverage and reasonable distribution are reserved in the archive. The mixed strategy skillfully balances the efficiency in the algorithm process and the quality of the final solution set, avoids the substantial increase of the overall calculation cost, and the diversity advantage makes up for the defects of solution set convergence under single strategy, so that a solution set meeting the Pareto optimality and sufficient diversity can be finally output, which perfectly adapts to the actual demand of rich optional solutions in multi-scheme decision-making in water, light and scenery multi-objective scheduling and other scenes. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The flow chart of the water, light and scenery optimization scheduling method based on the two-stage double-population evolution algorithm of the embodiment of the application.

[0041] Figure 2 The pareto front comparison graph of the multiple algorithms of the embodiment of the application. DETAILED DESCRIPTION

[0042] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. In addition, the technical features involved in each embodiment of the present application described below can be combined with each other as long as they do not conflict with each other.

[0043] The present application is directed to the water, wind and light multi-objective scheduling problem, and proposes a two-stage dual-population evolutionary algorithm (TS-DPEA). Based on the two-stage dual-population framework, the adaptive operator selection strategy is the core, the multi-source elite migration is the support, and the hybrid environment selection is the guarantee. The systematic design can realize the multi-objective optimization scheduling of water, wind and light. Specifically as follows:

[0044] (1) Dual-population-two-stage-adaptive coordination framework

[0045] Unlike traditional multi-objective algorithms using a single population, the evolutionary system is designed by two parallel populations (Pop1, Pop2), which breaks through the traditional single population evolution mode. Through the dynamic switching from fixed division exploration stage to adaptive coordination development stage, the global exploration and local exploitation ability are balanced, and the contradiction between convergence and diversity in multi-objective optimization is solved. Specifically as follows:

[0046] 1) Population diversity

[0047] Population diversity is used to quantify the dispersion degree of individuals in the objective space, and is one of the core judgment indexes for stage switching, which can reflect the exploration ability of the algorithm in the early stage. For two populations (denoted as population 1 and population 2), and each individual contains two objectives, the diversity of the two populations needs to be calculated first, and then the average of the two is taken as the overall diversity index. Specifically as follows:

[0048] Diversity of population 1:

[0049]

[0050] Diversity of population 2:

[0051]

[0052] The overall diversity index takes the average of the diversity of the two populations:

[0053]

[0054] In the formula: The sizes of population 1 and population 2, respectively; , These are the diversity values ​​for population 1 and population 2, respectively. The average diversity of the two populations reflects the overall degree of dispersion. The larger the value, the higher the degree of dispersion of the population and the stronger its exploration ability. Describe the objective function for the i-th individual; This is the Euclidean distance operator, used to calculate the spatial distance between two target vectors, for example... , , These are the first and second objective function values ​​of individual i in population 1, respectively.

[0055] 2) Phase switching conditions

[0056] The algorithm automatically triggers stage switching based on a diversity threshold or the percentage of function evaluations. The switching logic formula is as follows:

[0057]

[0058] In the formula: The numbers represent stages, with 0 indicating a fixed division of labor and 1 indicating an adaptive collaboration stage. FE is the diversity threshold; FE is the number of algorithm evaluations; maxFE is the maximum number of evaluations. This is the threshold for the percentage of evaluations.

[0059] 3) Two-population evolutionary model

[0060] The two populations use different operator rules at different stages to achieve a combination of division of labor and cooperation. The GA operator and DE operator are preferred as examples for further explanation, and the formula is described as follows:

[0061] (5)

[0062] In the formula: For the k-th population ( GA stands for Genetic Algorithm operator, and DE stands for Differential Evolution Algorithm operator. Both can maintain population diversity and prevent the population from getting trapped in local optima. , Let represent the selection probabilities of GA and DE during the adaptive cooperative phase.

[0063] Phase 1 (Fixed Division of Labor Phase): The two populations initially adopt a fixed division of labor to conduct extensive exploration. The two populations use the GA and DE operators respectively, making full use of the characteristics of different operators to conduct global exploration, aiming to quickly locate promising areas and maintain a high level of diversity.

[0064] Second stage (adaptive coordination stage): when the population diversity is lower than the threshold or the evaluation times reach a certain proportion, the algorithm automatically switches to this stage, dynamically switches to the adaptive coordination mode according to the population diversity, dynamically mixes GA and DE operators in the two populations according to the real-time performance, combines the global exploration advantage of GA and the local fine search advantage of DE of continuous variables (such as water level), and selects the operator through the success rate of the operator based on the selection strategy under different environments.

[0065] The double-population, two-stage and adaptive coordination framework avoids the defects of a single algorithm, ensures the sufficiency of global exploration in the early stage, and has adaptability and high efficiency of local exploitation in the later stage. The mechanism ensures the dual reinforcement and effective connection of global exploration ability and local development ability from the system architecture level, and provides a structural basis for solving the contradiction between convergence and diversity.

[0066] (2) Dynamic adaptive mechanism for operator performance

[0067] Traditional multi-operator algorithms mostly use fixed probability or random method to select operators, which cannot dynamically adjust according to the real-time performance of the optimization process and the operator, and the search efficiency is low. The application proposes a dynamic operator selection strategy based on the success rate of historical different environment screening, dynamically adjusts the selection probability of the operator, the higher the proportion of the operator in producing high-quality offspring and being reserved, the higher the score, and the greater the probability of being selected in the future, so that the algorithm can learn the optimal operator in the current optimization stage, and improve the evolution efficiency. The proportion of the offspring of the operator being reserved by the next generation after the offspring is generated is regarded as the current success rate of the operator, and the score is updated by a smoothing mechanism. The specific steps are as follows:

[0068] 1) Operator success rate calculation

[0069] The operator success rate is defined as the proportion of the offspring generated by the operator being reserved by the next generation population, and the calculation formula is:

[0070]

[0071] In the formula: is the success rate of the tth generation operator; is the operator type; t is the current iteration number; is the total number of offspring generated by the tth generation operator; is the number of offspring generated by the tth generation operator and reserved by the environmental selection strategy of the tth generation.

[0072] 2) Operator score update

[0073] In order to avoid the influence of single success rate fluctuation on selection, and at the same time retain the overall operator performance of history, the operator score is updated by a decay mechanism, and the formula is:

[0074] (7)

[0075] In the formula: The score of the tth generation operator The initial value ; is a decay coefficient, used to balance the weight of historical scores and current success rate.

[0076] 3) Operator selection probability calculation

[0077] According to the operator score, the selection probability is dynamically allocated, and the operator with higher performance obtains higher selection opportunity, so as to realize the resource inclination to the high-efficiency operator and solve the low-efficiency resource waste problem of the traditional fixed probability selection. The calculation formula is:

[0078]

[0079] In the formula: The selection probability of the tth generation operator , and .

[0080] (3) Multi-level and multi-source elite migration strategy

[0081] Most of the co-evolutionary algorithms have limitations, only focusing on the exchange of information between populations, but ignoring the guiding role of historical elite solutions, resulting in insufficient information utilization. In order to break through this limitation, the invention designs a two-way, multi-source, multi-level elite migration mechanism to realize the organic combination of horizontal migration between populations and vertical migration of external archive (Archive), and after each population is updated in the co-adaptive stage, the top 10% of elite solutions are selected from the other population, and the top 10% of historical elite solutions are selected from the external archive (Archive), to realize the fusion of multi-source elites, and then the two types of elite solutions are migrated to the current population and participate in the next environment selection. This mechanism not only promotes the horizontal communication between populations and accelerates the spread of excellent patterns and genes between populations, but also realizes the vertical inheritance of historical elites, effectively avoiding the problem of extinction of excellent genes due to the population falling into local optimum. Ultimately, not only the convergence speed of the algorithm is significantly improved, but also the population diversity is enriched, and the robustness and global search ability of the algorithm are enhanced. The specific steps are as follows:

[0082] 1) Elite migration quantity

[0083] The number of elite migration should consider both the effectiveness of information transmission and the stability of the population. The number of elites selected from each migration source is 10% of the total number of current effective individuals in the migration source. This proportion is determined based on the characteristics of the solution space of the water, wind and light scheduling problem. The 10% proportion can ensure that the migrated elites carry enough high-quality information (such as high power generation scheduling schemes explored by different populations and low residual load fluctuation schemes preserved in the archive), and also will not interfere with the evolution direction of the current population due to the excessive number of migrated individuals. The 10% proportion calculation result may be a decimal number, which needs to be rounded up.

[0084] 2) Sorting rules for elite selection

[0085] The elite selection of the two types of migration sources follows the dual standards of non-dominated priority and dispersion priority, to ensure the convergence of the elite:

[0086]

[0087] In the formula: is the non-dominated ranking level of the individual in the migration source X; is the crowding distance of the individual in the migration source X; First, sort by in ascending order, and then sort by in descending order.

[0088] 3) Construction and merging of migration elite set

[0089] Merge the elite of different populations and the archive elite to form the final migration set, the formula is as follows:

[0090]

[0091] In the formula: The elite subset extracted from the different populations realizes horizontal migration and transmits real-time optimization information; The elite subset extracted from the external archive realizes vertical migration and transmits historical high-quality information; The final elite set for migration into the current population will be merged with the offspring of the current population after environmental selection and participate in the evolution of the next round of iteration.

[0092] (4) Mixed environmental selection strategy

[0093] To solve the problem that the single environment selection strategy is difficult to comprehensively consider the calculation efficiency and solution set quality, the application designs a mixed selection system of different scenes and differentiation, which respectively adapts the core advantages of the two strategies to different scenes. For the external archive update of each generation, the fast non-dominated sorting and crowding distance strategy which is high in calculation efficiency is adopted to preferentially ensure the algorithm iteration speed and meet the timeliness demand of water, wind and light scheduling. For the maintenance of the external archive, the Pareto intensity calculation and minimum distance truncation strategy is triggered regularly to utilize the double advantages of diversity and convergence to accurately maintain the uniformity and coverage of the historical elite solution distribution, so as to ensure that the archive retains high-quality solutions with wide coverage and reasonable distribution. The mixed strategy skillfully balances the efficiency and final solution set quality in the algorithm process, avoids a large increase in the overall calculation cost, and compensates for the defects of solution set convergence under a single strategy with the advantage of diversity. Finally, a solution set meeting the Pareto optimality and sufficient diversity can be output, which perfectly adapts to the actual demand of rich optional solutions in multi-scheme decision-making in water, wind and light multi-objective scheduling and other scenes. The specific steps are as follows:

[0094] The external archive Archive is updated every K generations by using the environment selection of Pareto intensity and Euclidean distance truncation, and the rest of the time is selected by using fast non-dominated sorting and crowding distance.

[0095] 1) The environment selection of non-dominated sorting layering and crowding distance screening can realize efficient population screening and ensure that the calculation overhead of each generation population iteration is controllable. The individuals in the population are divided into different frontiers FrontNo according to the objective function, and the smaller FrontNo is, the higher the non-dominated level of the individual is. The individuals are screened according to the non-dominated level first, and then screened according to the crowding distance. The crowding distance is the sparseness of the individual in its own frontier, which avoids solution set convergence, and the formula is:

[0096]

[0097] In the formula, m is the number of objective functions; is the maximum and minimum value of the mth objective in the current frontier; is the target value of the adjacent individual of the ith individual after sorting in the mth objective.

[0098] Specifically, the non-dominated sorting is first performed on the current population to obtain a plurality of non-dominated layers, the non-dominated level of the first non-dominated layer is the highest, and the domination level of the subsequent layers is sequentially decreased. According to the priority of the non-dominated layers from high to low, the individuals of each layer are sequentially included in the external archive: after being included in the current layer, if the cumulative number of individuals in the archive does not reach the archive size, the individuals of the next layer are continuously included, and the process is recursively performed; if the cumulative number of individuals in the archive exceeds the archive size after being included in a certain layer, the individuals of the subsequent layers are stopped from being included, and only the individuals of the layer are calculated for the crowding distance. The individuals of the layer are sorted in descending order according to the crowding distance, and the first k individuals (k is the size of the archive size minus the total number of individuals included in all previous layers) are selected to supplement the external archive, so that the number of individuals in the archive is exactly equal to the archive size. If all the non-dominated layer individuals are included, the cumulative number still does not reach the archive size, and all the included individuals are directly stored in the external archive (generally, the population size before screening is definitely greater than the archive size).

[0099] 2) The distribution of the Archive solution set is optimized by Pareto intensity calculation and minimum distance truncation. The Pareto intensity reflects the comprehensive domination relationship and the proximity distance, and quantifies the individual superiority. The calculation formula is:

[0100]

[0101] In the formula: represents the degree of being dominated by high-quality individuals in the current Archive, the smaller the value, the lower the degree of being dominated by the individual; represents the degree of isolation of the individual in the target space, is the Euclidean distance between the neighboring individuals, the smaller the value, the more isolated the individual, and the solution set is avoided to be concentrated.

[0102] Specifically, first, the Pareto intensity value is used for preliminary screening to screen out all individuals in the population with a Pareto intensity value less than 1. If the number of screened individuals does not exceed the archive size, all qualified individuals are directly stored in the external archive. If the archive size is exceeded, the Euclidean distance between each two individuals is calculated and a distance matrix Dist is constructed, then the minimum proximity distance is used as the crowding degree judgment index, the crowding degree of each individual is calculated based on the matrix, and the most crowded individual is locked and deleted. Then, the Euclidean distance between each two individuals is recalculated, the distance matrix is updated, and the process of calculating the crowding degree, locking and deleting the most crowded individual is repeated to maintain the uniformity of the solution set, perform truncation selection, and finally store these individuals in the external archive.

[0103] Based on the TS-DPEA algorithm, the water, wind and light optimal scheduling method based on the two-stage double population evolutionary algorithm is provided, the reservoir water level is taken as the decision variable, the maximum power generation of the water, wind and light multi-energy complementary system and the minimum residual load mean square deviation are taken as the objective function, and each constraint is set to construct a water, wind and light multi-objective scheduling model; the water, wind and light multi-objective scheduling model is optimized and solved, as shown in the following formula: Figure 1 The method comprises the following steps:

[0104] S1: parameter and population initialization

[0105] The core parameters of the algorithm are set, the size of a single population N is 100, the maximum evaluation number maxFE, the size of the external archive (Archive) is 1000, the population diversity threshold is 1e-4, the evaluation number ratio threshold is 0.2, the operator score attenuation coefficient is 0.9, the archive is initialized, the elite solution in the iteration process is stored, and the initial value is empty; and two parallel initial populations (Pop1 and Pop2) are generated.

[0106] S2: first stage-fixed division global exploration

[0107] According to the population evolution rule, in the first stage, Pop1 is fixed to use the genetic algorithm operator (GA) to play the global search characteristic, and Pop2 is fixed to use the differential evolution algorithm operator (DE) to play the local search advantage of the water level continuous variable, and the wide area exploration is realized through the characteristic difference of the two types of operators; the offspring generation and screening are performed, the offspring individuals are generated based on the GA / DE operator, and the infeasible solution is removed in combination with the water, wind and light scheduling constraints (such as water balance, upper and lower limits of hydropower station output); the population and the archive are updated: the fast non-dominated sorting and crowdedness distance environment selection strategy is used to update the population, and the high-quality solution is reserved; meanwhile, the non-dominated solution generated in the iteration is stored in the external archive, 1 iteration is completed, and S3 is entered.

[0108] S3: stage switching condition judgment

[0109] The population diversity is calculated, and the individual dispersion degree of the target space is quantified; if any of the following conditions is met, the stage switching is triggered, and the second stage is entered: ① the population diversity is less than the set threshold; ② the evaluation number ratio (FE / maxFE) is greater than the evaluation number ratio threshold; if not, return to S2 to continue fixed division exploration.

[0110] S4: second stage-adaptive cooperative local exploitation

[0111] The success rate of the GA / DE operator is calculated according to the proportion of offspring generated by the GA / DE operator in the t-th generation, that is, the proportion of the number of offspring retained to the total number of offspring; the GA / DE operator score is updated using a decay mechanism to smooth the update and retain historical overall scores to avoid the influence of single fluctuations; the selection probability is calculated according to the operator score, which satisfies the probability sum of 1, and Pop1 and Pop2 are dynamically selected to generate offspring by the GA / DE operator to combine the advantages of the two types of operators.

[0112] Determining the number of migrations: the number of migrated individuals is calculated according to 10% of the population size and the size of the external archive, and multi-level elitist migration is performed; screening elite solutions: screening the top 10% elite solutions from the heterogeneous population for horizontal migration to transfer real-time optimization information, and screening the top 10% historical elite solutions from the external archive for vertical migration to transfer historical high-quality information, and merging into a migration elite set; merging the migration elite set with the offspring of the current population to participate in the subsequent environmental selection of the population.

[0113] Archive maintenance: K=5 can ensure the uniformity of the archive solution distribution while avoiding the efficiency loss caused by frequent calculations, so the external archive is updated every 5 generations using the Pareto intensity calculation and Euclidean distance truncation environmental selection strategy, and the fast non-dominated sorting and crowding distance environmental selection strategy is still used in other iterations to ensure the iteration efficiency of the algorithm, improve the convergence and distribution of elite solutions, complete one iteration, and return to S3 until the termination condition is met.

[0114] S5: Iteration termination and optimization result output

[0115] If the number of algorithm evaluations reaches maxFE, the iteration is terminated; the non-inferior solutions are extracted from the external archive to form a Pareto front, and the water, wind and light multi-objective scheduling optimization scheme set is output, covering the maximum power generation, minimum residual load variance and compromise multiple scheduling schemes for decision selection.

[0116] The following is a specific embodiment:

[0117] Taking the Baihetan Power Station in 2022 as the research object, a multi-objective scheduling model is constructed for optimization, and the optimization objectives are the maximum power generation on the power supply side and the minimum residual load variance on the power grid side. The dry season (strong wind and strong light in March 2022) of the Baihetan Hydropower Station is selected as the research scenario to carry out multi-objective optimization and scheduling research of the water, wind and light complementary system. The water level of the Baihetan Hydropower Station in early March 2022 is 788.91m, and the end water level is 795.75m, with an average inflow of 1790.32m³ / s; in this example, the Baihetan Hydropower Station output is sent to Zhejiang Province at a proportion of 50%, and the average monthly load of Zhejiang Province in March 2022 is about 60081MW.

[0118] (1) Model construction

[0119] Using water level as the decision variable, a multi-objective function model is established with the objectives of maximizing the power generation of the hydro-wind-solar multi-energy complementary system and minimizing the mean square error of the surplus load. Constraints such as water level, flow rate, and output are set to jointly constitute a multi-objective scheduling model.

[0120] The power output of the hydropower station in each time period is calculated based on the net head and water consumption rate curve of the hydropower station in each time period. At the same time, the wind and solar power output in the corresponding time period is added. The first objective function is established with the combined power output of hydropower, wind and solar power in each time period, with the goal of maximizing the power generation of the hydropower-wind-solar multi-energy complementary system:

[0121] In the formula: The total power generation of the complementary system is represented by T, the total number of time periods within the scheduling period is represented by T, and t is the time period sequence number. , It is the scheduling period. It is the total output during time period t. This represents the power output of the hydropower station during time period t. and These represent the power output of wind power stations and photovoltaic power stations during the same time period, respectively.

[0122] Meanwhile, in addition to considering the power generation benefits on the power supply side, the power station should also consider the peak-shaving effect on the grid side. Therefore, a second objective function was established based on the load demand of the power station's transmission area and the total output process of the hydro-wind-solar system:

[0123]

[0124] In the formula: This represents the mean square error of the residual load of the power grid after being supplied by a multi-energy complementary system of water, wind, and solar power. Indicates the power grid load demand. This indicates the surplus load of the power grid after being supplied by the hydro-wind-solar multi-energy complementary system. This represents the average surplus load of the power grid.

[0125] Set model constraints for the multi-objective scheduling model. These constraints include: reservoir water level constraints, water balance constraints, reservoir characteristic constraints, power output characteristic constraints, flow rate constraints, water level / flow rate variation constraints, hydropower station power output constraints, wind farm wind speed constraints, and photovoltaic power generation capacity constraints. Specific constraint settings are as follows: 1) Reservoir water level constraints.

[0126]

[0127] In the formula: Indicates that the reservoir is Water level in front of the dam during the specified time period and Reservoirs at The upper and lower limits of the water level in front of the dam during a certain period.

[0128] 2) Water balance constraint

[0129]

[0130] where: and are the reservoir storage at time period t and t-1, respectively; is the inflow to the reservoir at time period t; is the outflow from the reservoir at time period t.

[0131] 3) Reservoir characteristics constraint

[0132]

[0133] where: is the average upstream water level of the reservoir at time period t; is the average downstream tailwater level of the reservoir at time period t; , are the reservoir water level-storage curve and tailwater-outflow curve, respectively.

[0134] 4) Power output characteristics constraint

[0135]

[0136] where: is the reservoir power output calculation function, is the net head of the reservoir t at time period t; is the power reference flow of the reservoir t at time period t.

[0137] 5) Flow constraint

[0138]

[0139] where: and are the lower and upper limits of the outflow of the reservoir t at time period t.

[0140] 6) Water level / flow variation constraint

[0141]

[0142] where: is the allowed variation interval of the water level within the reservoir scheduling time period interval.

[0143] 7) Hydropower station power output constraint

[0144]

[0145] In the formula: and These respectively represent the reservoir at t The upper and lower limits of output during different time periods.

[0146] 8) Wind speed constraints at wind farms

[0147]

[0148] In the formula: and These respectively represent wind farms in t Upper and lower limits of wind speed for entry and exit points during specific time periods.

[0149] 9) Capacity constraints of photovoltaic panels

[0150]

[0151] In the formula: and They represent the photoelectric field at... t Upper and lower limits of the capacity of the power generation panels during different time periods.

[0152] This invention employs the TS-DPEA algorithm for solving the problem and selects classic multi-objective optimization algorithms NSGAⅡ, MOBCA, SPEA2, and MOEA / D as comparative algorithms. The parameters of the TS-DPEA algorithm are set as follows: population size N=100, maximum number of iterations MaxIter=1000, external archive Archive=100; all other algorithm parameters are set to default.

[0153] After algorithm optimization, each algorithm retains 100 solutions to form the Pareto front solution. Table 1 shows the multi-objective scheduling schemes obtained by TS-DPEA optimization:

[0154] Table 1 Optimization Results of TS-DPEA Algorithm

[0155]

[0156] As shown in Table 1, when the system's power generation increases, the mean square error of the remaining load also increases, and the peak-shaving efficiency decreases, indicating a significant competitive relationship between power generation and the mean square error of the remaining load. Therefore, the non-dominated solution set obtained from the example can effectively reflect the game between the economic benefits on the source side and the peak-shaving demand on the grid side, which is consistent with engineering practice.

[0157] Pareto fronts of various algorithms, for example Figure 2As shown, it can be seen that all have reached the convergence state, and the Pareto front of the TS-DPEA algorithm in the power generation and residual load variance dual target space, when the residual load variance is at a lower level, the TS-DPEA can achieve significantly higher power generation; and with the increase of power generation, the growth of residual load variance is more gentle, the trade-off efficiency between the two targets is better, and the value range of TS-DPEA on the two target functions is larger, providing more rich scheduling schemes for subsequent decision-making.

[0158] In contrast, the Pareto front of SPEA2, MOBCA and NSGAII is more backward, the power generation is lower under the same residual load, or the residual load variance is higher under the same power generation; although MOEA / D pursues high power generation, the rising amplitude of residual load variance is larger, and the balance ability of dual targets is weaker than TS-DPEA, and the scheduling scheme is limited, which is difficult to meet the demand of actual scheduling scheme.

[0159] As can be seen, the TS-DPEA proposed in the application has more prominent trade-off performance in the dual target optimization of power generation improvement and residual load fluctuation control, and can provide more diverse scheduling schemes.

[0160] From different angles, the characteristics of the data are evaluated, the hypervolume index (HV) reflects the volume size covered by the non-dominated solution set in the target space, and the larger the value is, the better the performance of the solution set is; the distribution uniformity index (ΔP) measures the uniformity of the distribution of the solution set in the target space, and the smaller the value is, the more uniform the distribution is; the diversity index (DM) reflects the diversity of the solution set in each target dimension, and the larger the value is, the higher the diversity is.

[0161] As shown in Table 2, the hypervolume index of the TS-DPEA algorithm proposed in the application reaches 149830, which is 2.3 times the best value in the comparison algorithm, which directly proves that the application has achieved a synergistic improvement in the convergence and diversity of the solution set through systematic improvement, and the technical effect is remarkable; the diversity index of TS-DPEA is much higher than that of all comparison algorithms, which shows that it effectively avoids premature convergence, explores a wider target space, and generates more diversified scheduling schemes, perfectly meeting the urgent demand for multi-scheme decision-making in engineering practice. This leap in performance is the unexpected technical effect produced by the synergistic effect of the various technical features of the application. Although the distribution uniformity is not optimal, as a whole, it provides more diverse and high-quality schemes for water, wind and light multi-objective optimization scheduling decision-making scenarios, and is more competitive in water, wind and light multi-objective optimization problems.

[0162] Table 2 Comparison of evaluation indexes of various algorithms

[0163]

[0164] In summary, the present application provides a foundation through a dual-population architecture, realizes intelligent search through adaptive operator selection, promotes historical information cooperation and real-time information updating through elite migration, balances efficiency and quality through mixed environment selection, and is finally successfully applied to a complex water-wind-solar complementary optimization scheduling problem, thereby providing a water-wind-solar complementary system with a Pareto front scheduling scheme that has high convergence and high diversity and is generated efficiently and with high quality.

[0165] Those skilled in the art will easily understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application, and any modifications, equivalent replacements and improvements made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A water, wind, and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm, characterized in that, Includes the following steps: Using reservoir water level as the decision variable and the objective functions of maximizing power generation and minimizing the mean square error of surplus load in the hydro-wind-solar multi-energy complementary system, a multi-objective scheduling model for hydro-wind-solar is constructed. Two populations are initialized, and the multi-objective scheduling model for water, wind, and solar power is solved. Initially, both populations are in a fixed division of labor phase. When the phase switching condition is met, they switch to an adaptive cooperative phase for iterative updates, ultimately obtaining the solution set and achieving optimized scheduling of water, wind, and solar power. Wherein: Fixed division of labor phase: The two populations are iteratively updated using operators A and B respectively; where operator A and B are the genetic operator and the difference operator, respectively; Adaptive Coordination Phase: The selection probabilities of operators A and B are determined based on the real-time performance of the populations. The two populations then iteratively update their selections by choosing operator A or B based on these probabilities. Determining the selection probabilities of operators A and B based on the real-time performance of the populations includes: For operators A and B, the success rate of an operator is the proportion of offspring produced by the operator that are retained by the next generation population. Based on the operator success rate, an attenuation mechanism is used to update the operator score: in, t This represents the current iteration number. , The first t Alternative operators The rating and success rate For operator type, The attenuation coefficient; The probability of selecting an operator is determined by the proportion of its score to the sum of the scores of the two operators. In the adaptive cooperation phase, after each population is updated, the top 10% of elite solutions are selected from another population and the top 10% of elite solutions are selected from the external archive. Then, the union of the two types of elite solutions is migrated to the current population and participates in the next environment selection. The update method of the external archive is as follows: every K generations, individuals are selected from the population to update the external archive based on Pareto strength and Euclidean distance truncation. In the remaining generations, individuals are selected from the population to update the external archive based on non-dominance level and crowding distance. The phase switching condition is as follows: In the fixed division of labor phase, the population diversity is evaluated after each iteration update. When the population diversity is less than the diversity threshold or the number of iterations reaches the evaluation number threshold, the phase switching is performed.

2. The water, wind, and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm as described in claim 1, characterized in that, The method for determining population diversity is as follows: calculate the diversity of the two populations separately using the following formula, and then take the mean of the diversity of the two populations as the final population diversity. in, For the diversity of a single population, N is the population size. , Let be the objective function values ​​of the i-th and j-th individuals in the population, respectively. This is the Euclidean distance operator.

3. The water, wind, and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm as described in claim 1, characterized in that, The method for selecting elite solutions is as follows: first, individuals in the migration source are sorted in ascending order according to their non-dominance level; then, individuals in the migration source are sorted in descending order according to their crowding distance; and finally, the top 10% of individuals in the migration source after sorting are selected as elite solutions; the migration source is another group or an external archive.

4. The water, wind, and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm as described in claim 1, characterized in that, Individuals are selected from the population based on non-dominance level and crowding distance, specifically as follows: The population is sorted by non-dominated hierarchy to obtain several non-dominated layers ranked from high to low. Individuals in the non-dominated layers are sequentially placed into the external archive according to the ranking, stopping when the external archive reaches the archive size. After a non-dominated layer is included, if the cumulative number of individuals in the external archive does not reach the archive size, individuals from the next non-dominated layer are included, and so on. If the cumulative number of individuals in the external archive exceeds the archive size after a non-dominated layer is included, the inclusion of subsequent non-dominated layers is stopped. Only the crowding distance of all individuals in the non-dominated layer is calculated, and the individuals in the non-dominated layer are sorted in descending order of crowding distance. The top k individuals are selected to supplement the external archive, where k is the archive size minus the total number of individuals in all included non-dominated layers, so that the number of individuals in the external archive reaches the archive size.

5. The water, wind, and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm as described in claim 1, characterized in that, Individuals were selected from the population based on Pareto strength and Euclidean distance truncation, specifically as follows: Individuals in the population with a Pareto strength value less than 1 were selected; If the number of individuals selected does not exceed the archive size, all selected individuals will be directly stored in the external archive. If the number of selected individuals exceeds the archive size, first calculate the Euclidean distance between the selected individuals and construct a distance matrix. Based on this distance matrix, calculate the crowding degree of each individual and delete the most crowded individual. Then repeat the above process until the number of remaining individuals exactly matches the archive size, and store the remaining individuals in an external archive.

6. A water, wind, and solar energy optimization scheduling system based on a two-stage dual-population evolutionary algorithm, characterized in that, Includes a processor, the processor being configured to execute the water, wind and solar energy optimization scheduling method based on a two-stage dual-population evolutionary algorithm as described in any one of claims 1-5.

Images