A new energy power grid dispatching strategy economy optimization method and system

Abstract

This invention discloses a method and system for optimizing the economic efficiency of new energy power grid dispatching strategies, relating to the field of new energy power system dispatching technology. The method includes: acquiring historical source-load data, generating an initial scenario set and performing clustering and compression to obtain a test scenario library, and constructing an extreme scenario set; generating an initial strategy population and performing iterative evolution; simulating and calculating a comprehensive economic score in the test scenario library; identifying failed scenarios and extracting feature vectors; and performing directed evolutionary operations on the population using the comprehensive economic score and feature vectors; during the iteration process, staged stress tests are also conducted using the extreme scenario set, and the weights of extreme scenarios are dynamically adjusted based on the test results; and the final dispatching strategy is output after a preset termination condition is met. This invention can compress redundant test scenarios, mine test process information, effectively cover extreme scenarios, and significantly improve test efficiency, achieving synergistic optimization of the economic efficiency and security of dispatching strategies.

Description

Technical Field

[0001] This invention relates to an economic optimization method and system for new energy power grid dispatching strategies, belonging to the field of new energy power system dispatching technology. Background Technology

[0002] New energy power generation (wind power and photovoltaic) exhibits significant randomness and volatility, posing challenges to the safe and economical dispatch of the power grid. To evaluate the economics of different dispatch strategies, traditional methods require simulation testing across numerous new energy output scenarios. This involves calculating various cost indicators and then iteratively optimizing to find the optimal strategy. New energy output scenarios typically contain hundreds or thousands of time-series samples, many of which share similar characteristics. Performing complete simulations across all scenarios not only wastes computational resources and results in long testing cycles but also fails to meet the timeliness requirements of daily rolling optimization. Furthermore, traditional methods test before optimizing, with the testing phase only outputting economic indicators, leaving subsequent optimization directions unclear and convergence slow. Conventional random sampling methods also struggle to effectively cover high-impact, low-probability extreme scenarios (such as consecutive windless days or sudden heavy loads), leading to a sharp drop in the economics of the optimized strategy under extreme conditions, or even resulting in load shedding. Therefore, there is an urgent need for an economic optimization method for new energy power grid dispatch strategies that can compress redundant test scenarios, extract information from the testing process, and effectively cover extreme scenarios. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for optimizing the economic efficiency of new energy power grid dispatching strategies, aiming to solve the technical problems in the prior art, such as resource waste caused by a large number of test scenarios, difficulty in meeting timeliness requirements, slow convergence speed due to unclear optimization direction, and load loss caused by a sharp drop in strategy economic efficiency in extreme scenarios.

[0004] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0005] On the one hand, this invention provides an economic optimization method for new energy power grid dispatching strategies, comprising:

[0006] Historical source load data is acquired, an initial scene set is generated and clustered and compressed to obtain a test scene library; and an extreme scene set is constructed based on historical source load data and preset filtering rules.

[0007] Multiple scheduling policies are generated to form an initial policy population, and the following iterative evolutionary process is performed:

[0008] Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy;

[0009] Based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios.

[0010] By using a comprehensive economic score and feature vectors, a directed evolutionary operation is performed on the current strategy population to generate a new generation of strategy population;

[0011] During the iterative evolution process, a set of extreme scenarios is used to conduct phased stress tests on the strategies in the iteration, and the weight of extreme scenarios in subsequent optimizations is dynamically adjusted based on the stress test results.

[0012] When the preset termination conditions are met, the final scheduling strategy is output.

[0013] Optionally, the initial scene set is clustered and compressed to obtain a test scene library, including:

[0014] The K-means clustering algorithm was used to cluster the initial scene set. The features used for clustering included daily average power output, daily power output fluctuation rate, daily power output peak-valley difference, and the correlation between daily power output and load.

[0015] The central scene of each cluster category is selected as a representative scene, and the selected representative scenes constitute the test scene library;

[0016] The proportion of the initial scenes contained in each cluster category to the total number of initial scenes is used as the scene weight of the representative scene in that category.

[0017] Optionally, multiple scheduling policies are generated to form an initial policy population, including:

[0018] Set the scheduling strategy parameters to be optimized, including: minimum output coefficient of thermal power, energy storage charging threshold, energy storage discharging threshold, and wind and solar curtailment tolerance.

[0019] Randomly sample the scheduling policy parameters to generate N scheduling policies, forming an initial policy population.

[0020] Optionally, simulations are performed on each scheduling policy in the current policy population within the test scenario library to calculate the comprehensive economic score of each scheduling policy. For any given scheduling policy, the following operations are performed:

[0021] In each test scenario, the scheduling strategy is simulated to obtain a single-scenario economic score:

[0022] ;

[0023] in, Total operating cost, Losses due to wind and solar power curtailment For carbon emission costs, To incur penalties for load loss, , , , Let be the weight coefficient, and satisfy... ;

[0024] Based on the scenario weights, the single-scenario economic scores of each test scenario are weighted and summed to obtain the comprehensive economic score of the scheduling strategy.

[0025] Optionally, based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios, including:

[0026] Select scheduling strategies with an overall economic score lower than a preset threshold, identify the k worst single-scenario economic scores of the selected strategy in each scenario in the test scenario library, and mark them as failed scenarios.

[0027] Extract feature vectors for failure scenarios, including: new energy output volatility, net load ramp-up rate, and new energy penetration rate.

[0028] Optionally, using a comprehensive economic score and eigenvectors, a directed evolutionary operation is performed on the current policy population to generate a new generation of policy populations, including:

[0029] The current strategy population is ranked according to the comprehensive economic score, and the M strategies with the highest scores are retained as elite strategies.

[0030] By performing crossover and mutation operations on elite strategies, a new generation of strategy populations can be generated;

[0031] Specifically, for the new strategy generated by the mutation operation, its neighboring domain in the feature vector space of the failure scenario is detected; if the failure scenario exists in the neighboring domain, the parameters of the new strategy are adjusted in a targeted manner, and the targeted adjustment includes at least one of the following: increasing the minimum output coefficient of thermal power, increasing the energy storage discharge threshold, decreasing the energy storage charging threshold, and decreasing the tolerance for wind and solar curtailment.

[0032] Optionally, the strategy in the iteration can be subjected to periodic stress tests using a set of extreme scenarios, and the weights of the extreme scenarios in subsequent optimizations can be dynamically adjusted based on the stress test results, including:

[0033] After each Z iterations of optimization, the strategy with the best overall economic score in the current strategy population will be specifically tested in the set of extreme scenarios.

[0034] If the economic performance score of the optimal strategy in a single scenario is lower than the preset threshold in a certain extreme scenario, the proportion of that extreme scenario will be increased in the next special test.

[0035] Optionally, based on historical source load data and preset filtering rules, an extreme scenario set is constructed, which includes at least one of the following scenarios:

[0036] Scenarios where the daily output of new energy sources is less than 10% of the installed capacity, scenarios where the hourly output fluctuation rate of new energy sources exceeds 50%, and peak load scenarios.

[0037] Optionally, iteration stops when any of the following preset termination conditions are met:

[0038] The preset maximum number of iterations has been reached;

[0039] Alternatively, the overall economic score of the current strategy population converges in the test scenario library, and in the most recent phased stress test, the single-scenario economic score of the optimal strategy in each extreme scenario is greater than the preset threshold.

[0040] Secondly, the present invention provides an economic optimization system for new energy power grid dispatching strategies, comprising:

[0041] The scenario construction module is used to acquire historical source load data, generate an initial scenario set, and cluster and compress it to obtain a test scenario library; and to construct an extreme scenario set based on historical source load data and preset filtering rules.

[0042] The initial policy generation module is used to generate multiple scheduling policies, forming an initial policy population;

[0043] The iterative evolution module is used to perform the following iterative evolution process:

[0044] Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy;

[0045] Based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios.

[0046] By using a comprehensive economic score and feature vectors, a directed evolutionary operation is performed on the current strategy population to generate a new generation of strategy population;

[0047] The stress test module is used to conduct phased stress tests on the strategy during the iterative evolution process using a set of extreme scenarios, and dynamically adjust the weight of extreme scenarios in subsequent optimizations based on the stress test results.

[0048] The strategy output module is used to output the final scheduling strategy when the preset termination conditions are met.

[0049] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0050] This invention compresses numerous scenarios through clustering, reducing simulation computation and meeting the timeliness requirements of intraday rolling optimization. During optimization, it performs targeted evolutionary operations on the current policy population based on the feature vectors of failed scenarios, quantifying weaknesses exposed during testing into optimization directions, guiding targeted policy enhancement, avoiding blind searches, and improving optimization convergence speed. It employs phased testing in extreme scenarios to ensure the policy maintains economy and security even under extreme conditions. Furthermore, this invention requires only clustering and basic evolutionary algorithms, eliminating the need for complex deep learning models and reinforcement learning frameworks, resulting in low computational overhead and easy deployment. The final output policy can be simplified to scheduling policy parameters, easily scheduling any understanding and decision logic, improving the acceptability of practical applications. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the economic optimization method for new energy power grid dispatching strategy provided in an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of scene compression and weight allocation provided in an embodiment of the present invention. Detailed Implementation

[0053] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0054] Example 1

[0055] This embodiment introduces an economic optimization method for new energy power grid dispatching strategies, such as... Figure 1 As shown, it specifically includes the following:

[0056] S1. Obtain historical source load data, generate an initial scene set, and cluster and compress it to obtain a test scene library; and, based on historical source load data and preset filtering rules, construct an extreme scene set;

[0057] We collected wind power and solar power output and load data from a major energy base over the past year, collecting one point every 15 minutes, for a total of 35,040 time-series points over the year. We then generated 365 daily scenarios on a daily basis as the initial scenario set.

[0058] like Figure 2As shown, this step uses the K-means clustering algorithm to cluster 365 daily scenarios. The features used for clustering include daily average power output, daily power output fluctuation rate, daily power output peak-valley difference, and the correlation between daily power output and load. The correlation between daily power output and load is represented by the Pearson correlation coefficient.

[0059] Specifically, the number of clusters K=10 is set, and the 365 daily scenes are clustered into 10 classes. The scene closest to the centroid in each class is selected as the representative scene of that class. All representative scenes constitute the test scene library, that is, the test scene library in this example has 10 scenes. The proportion of the initial scene count in each cluster to the total number of initial scenes is used as the scene weight of the representative scene of that class, that is, the weight of the test scene. Scene compression and scene weight allocation are shown in Table 1:

[0060] Table 1 Scene Compression and Weight Allocation Table

[0061]

[0062] Constructing an extreme scenario set: Based on historical wind power and photovoltaic power output data and load data, we extract and construct three scenarios: continuous low power output scenario where the daily power output of new energy is less than 10% of the installed capacity, scenario where the hourly power output fluctuation rate of new energy exceeds 50%, and load peak scenario. These three scenarios constitute the extreme scenario set of this embodiment.

[0063] S2. Generate multiple scheduling policies to form an initial policy population, and perform iterative evolution;

[0064] Specifically, the scheduling strategy parameters to be optimized are set, including:

[0065] Minimum output coefficient of thermal power ;

[0066] Energy storage charging threshold Charging occurs when the output of new energy sources exceeds the load ratio;

[0067] Energy storage discharge threshold Discharge occurs when the output of new energy sources is insufficient to meet the load ratio;

[0068] Tolerance for wind and solar power curtailment The upper limit for the proportion of new energy sources that should be discarded when they exceed the capacity to absorb them.

[0069] Randomly sample the scheduling policy parameters to generate N=50 scheduling policies, forming the initial policy population.

[0070] The specific process of iterative evolution is as follows:

[0071] S2.1 Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy;

[0072] Specifically, the 50 scheduling policies were simulated on 10 representative test scenarios, with each scenario running for 24 hours and a time resolution of 15 minutes. For any given scheduling policy, the following operations were performed:

[0073] In each test scenario, the scheduling strategy is simulated to obtain a single-scenario economic score:

[0074] ;

[0075] in, Total operating cost, Losses due to wind and solar power curtailment For carbon emission costs, To incur penalties for load loss, , , , Let be the weight coefficient, and satisfy... In this example, , , , The comprehensive economic score S is used to quantify the overall economic performance of the power grid. The higher the score, the better the overall economic performance of the system.

[0076] Based on the weights of each test scenario, the single-scenario economic scores S for each test scenario are weighted and summed to obtain the comprehensive economic score of the scheduling strategy. .

[0077] Specifically, losses from wind and solar power curtailment Calculation: When the output of new energy sources Exceeding the system's capacity limit At this time, power wastage occurs; the upper limit of the system's acceptance capacity is determined by the electrical load, the minimum output of thermal power, and the maximum charging power of energy storage. Joint decision:

[0078] ;

[0079] For electrical load, This refers to the rated capacity of the thermal power unit.

[0080] ;

[0081] For the on-grid tariff of new energy, For time resolution. To contribute to new energy sources (wind power, photovoltaics) at time t. Let t be the upper limit of the system's capacity to accept data at time t.

[0082] In practice, the proportion of curtailed power is affected by the tolerance for wind and solar power curtailment. limit:

[0083] Among them, the theoretical power generation of new energy is the total amount of electricity that can be generated by new energy at full power generation within the assessment period.

[0084] Specifically, total system operating cost ; For the operating costs of thermal power units, Costs associated with energy storage charging and discharging losses. This refers to the startup cost of thermal power units.

[0085] Energy storage charge / discharge decisions are based on energy storage charging thresholds. and energy storage discharge threshold control:

[0086] when During charging, the energy storage power is:

[0087] ;

[0088] when At that time, the stored energy is discharged, and the discharge power is:

[0089] ;

[0090] in, Let t be the system's electrical load. This represents the maximum charging power of the energy storage system. This represents the maximum discharge power of the energy storage system.

[0091] Energy storage charging and discharging loss costs:

[0092] ;

[0093] in, For charging efficiency, For discharge efficiency; This is the energy storage cycle loss coefficient.

[0094] Operating costs of thermal power units Calculation:

[0095] The output of thermal power units is determined by power balance:

[0096] ;

[0097] Thermal power unit output is constrained by minimum output:

[0098] ;

[0099] in, Let t be the actual output of the thermal power unit.

[0100] The quadratic function coal consumption model is used, and the calculation formula is as follows:

[0101] ;

[0102] Where a, b, and c are the coal consumption characteristic parameters of thermal power units.

[0103] Carbon emission calculation cost calculate:

[0104] ,in For carbon trading prices, Carbon emission intensity of thermal power units.

[0105] Calculation of the cost P for loss of load:

[0106] ;

[0107] in, The penalty coefficient per unit of load loss. Let t be the system load loss at time t.

[0108] As can be seen from the above formula, the essence is still to optimize and adjust the minimum output coefficient of thermal power. Energy storage charging threshold Energy storage discharge threshold And tolerance for wind and solar power curtailment By changing the values ​​of each cost item and loss, the economic optimization of the scheduling strategy can be achieved.

[0109] S2.2, Based on comprehensive economic evaluation Select a subset of scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios;

[0110] Select scheduling strategies with an overall economic score lower than a preset threshold, identify the k worst test scenarios with the single-scenario economic score S in each scenario of the test scenario library for the selected strategy, and mark them as failed scenarios.

[0111] Extract feature vectors for failure scenarios, including: new energy output volatility, net load ramp-up rate, and new energy penetration rate;

[0112] Specifically, regarding the comprehensive economic score For each of the 10 strategies with k=10 values ​​below a preset threshold, identify the 3 scenarios with the worst economic performance in each scenario and extract the feature vectors of these scenarios, including: the volatility of new energy output, the net load ramp-up rate, and the penetration rate of new energy.

[0113] New energy output volatility: , The installed capacity is divided by the installed capacity to eliminate the impact of capacity differences;

[0114] raw standard deviation ; This represents the number of sampling points; The average output of new energy sources within the statistical period;

[0115] Net load ramp rate: ; Let the net load value at time t be defined as... ; For the previous moment Net load value, For time intervals;

[0116] New energy penetration rate: .

[0117] S2.3. Using the comprehensive economic score and feature vector, perform a directed evolution operation on the current strategy population to generate a new generation of strategy population;

[0118] The current strategy population is ranked according to the comprehensive economic score, and the top 10 strategies with the highest scores (M=10) are retained as elite strategies.

[0119] By performing crossover and mutation operations on elite strategies, a new generation of strategy populations can be generated;

[0120] Specifically, simulated binary crossover and polynomial mutation are performed to generate 40 new strategies, which together constitute a new generation of 50 strategies.

[0121] Real-time targeted enhancement during polynomial mutation operations: For a new strategy generated by the mutation operation, its neighboring domain in the feature vector space of the failure scenario is detected; if the failure scenario exists in the neighboring domain, the parameters of the new strategy are targetedly adjusted. This targeted adjustment includes at least one of the following: increasing the minimum output coefficient of thermal power, increasing the energy storage discharge threshold, decreasing the energy storage charging threshold, and reducing the tolerance for wind and solar curtailment. Specifically, this can involve shifting the parameters towards a more conservative direction, such as: Increase by 5%, Reduce by 5%, Increase by 5%, Reduced by 10%.

[0122] S3. During the iterative evolution process, extreme scenario sets are also used to conduct phased stress tests on the strategies in the iteration, and the weight of extreme scenarios in subsequent optimizations is dynamically adjusted based on the stress test results.

[0123] This embodiment constructs 15 extreme scenarios: 5 scenarios with continuous 72-hour low wind speeds (average wind power output <10%), 5 scenarios with daily photovoltaic volatility exceeding 50% (caused by rapidly moving clouds), and 5 scenarios with peak loads (summer high-temperature days with loads exceeding the forecast by 10%).

[0124] After each iteration of Z=5 rounds of optimization, the strategy with the best overall economic score in the current strategy population is specifically tested in 15 extreme scenarios:

[0125] If the economic score of the optimal strategy in a single scenario is lower than the preset threshold in a certain extreme scenario, the proportion of that extreme scenario will be increased in the next special test. For example, the weight will be increased to twice the original weight. In the first round of special tests, each extreme scenario will be configured with equal weight, which is 1 / 15.

[0126] S4. When the preset termination condition is met, output the final scheduling strategy.

[0127] The maximum number of iterations G is reached;

[0128] Or the overall economic score of the current strategy population converges in the test scenario library (e.g., after a set number of iterations). The value change rate is less than 2%, and in the most recent phased stress test, the single-scenario economic score of the optimal strategy in each extreme scenario is greater than the preset threshold. In the stress test, the calculation method of the single-scenario economic score is consistent with the calculation of the test scenario library.

[0129] If the overall economic score of the current strategy population has converged, but the economic score of the optimal strategy in some extreme scenarios does not reach the preset threshold, then continue iterative optimization, increase the weight of the extreme scenarios that have not met the threshold in subsequent iterations, and drive the strategy to optimize in a targeted manner; if the overall economic score remains converged for T consecutive iterations and the number of extreme scenarios that meet the threshold does not increase, then trigger forced termination, and output the scheduling strategy with the best overall economic score and the most extreme scenarios that meet the threshold during the iteration process.

[0130] The final output scheduling strategy includes the optimal strategy parameter, the minimum output coefficient of thermal power plants. Energy storage charging threshold Energy storage discharge threshold Tolerance for wind and solar power curtailment .

[0131] Example 2

[0132] Based on the same inventive concept as Embodiment 1, this embodiment provides an economic optimization system for new energy power grid dispatching strategies, comprising:

[0133] The scenario construction module is used to acquire historical source load data, generate an initial scenario set, and cluster and compress it to obtain a test scenario library; and to construct an extreme scenario set based on historical source load data and preset filtering rules.

[0134] The initial policy generation module is used to generate multiple scheduling policies, forming an initial policy population;

[0135] The iterative evolution module is used to perform the following iterative evolution process:

[0136] Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy;

[0137] Based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios.

[0138] By using a comprehensive economic score and feature vectors, a directed evolutionary operation is performed on the current strategy population to generate a new generation of strategy population;

[0139] The stress test module is used to conduct phased stress tests on the strategy during the iterative evolution process using a set of extreme scenarios, and dynamically adjust the weight of extreme scenarios in subsequent optimizations based on the stress test results.

[0140] The strategy output module is used to output the final scheduling strategy when the preset termination conditions are met.

[0141] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.

[0142] In summary, this invention compresses a large number of historical scenarios through clustering, reducing the computational load of simulations and meeting the timeliness requirements of intraday rolling optimization. During the optimization process, a directed evolutionary operation is performed on the current policy population based on the feature vectors of failed scenarios, quantifying the weaknesses exposed during testing into optimization directions, guiding the targeted enhancement of the policy, avoiding blind search, and improving the optimization convergence speed. Extreme scenario-based phased testing ensures that the policy maintains its economy and security even under extreme conditions. Moreover, this invention only requires clustering and basic evolutionary algorithms, without the need for complex deep learning models and reinforcement learning frameworks, resulting in low computational overhead and easy deployment. The final output policy can be simplified to scheduling policy parameters, easily scheduling any understanding and decision logic, improving the acceptability of practical applications.

[0143] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0145] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0147] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for optimizing the economy of a new energy power grid dispatching strategy, characterized in that, include: Acquire historical source load data, generate an initial scene set, and cluster and compress it to obtain a test scene library; In addition, an extreme scenario set is constructed based on historical source load data and preset filtering rules; Multiple scheduling policies are generated to form an initial policy population, and the following iterative evolutionary process is performed: Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy; Based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios. By using a comprehensive economic score and feature vectors, a directed evolutionary operation is performed on the current strategy population to generate a new generation of strategy population; During the iterative evolution process, a set of extreme scenarios is used to conduct phased stress tests on the strategies in the iteration, and the weight of extreme scenarios in subsequent optimizations is dynamically adjusted based on the stress test results. When the preset termination conditions are met, the final scheduling strategy is output. 2.The new energy grid dispatching strategy economy optimization method according to claim 1, characterized in that, The initial scene set is clustered and compressed to obtain a test scene library, including: The K-means clustering algorithm was used to cluster the initial scene set. The features used for clustering included daily average power output, daily power output fluctuation rate, daily power output peak-valley difference, and the correlation between daily power output and load. The central scene of each cluster category is selected as a representative scene, and the selected representative scenes constitute the test scene library; The proportion of the initial scenes contained in each cluster category to the total number of initial scenes is used as the scene weight of the representative scene in that category. 3.The new energy grid dispatching strategy economy optimization method according to claim 1, characterized in that, Multiple scheduling policies are generated to form an initial policy population, including: Set the scheduling strategy parameters to be optimized, including: minimum output coefficient of thermal power, energy storage charging threshold, energy storage discharging threshold, and wind and solar curtailment tolerance. Randomly sample the scheduling policy parameters to generate N scheduling policies, forming an initial policy population. 4.The method of claim 2, wherein, Simulate the scheduling strategies in the current strategy population within the test scenario library, and calculate the comprehensive economic score of each scheduling strategy. For any given scheduling strategy, perform the following operations: In each test scenario, the scheduling strategy is simulated to obtain a single-scenario economic score: ； wherein, is the total operation cost, is the wind and light curtailment loss, is the carbon emission cost, is the load loss penalty cost, , , , is a weight coefficient, and satisfies ; Based on the scenario weights, the single-scenario economic scores of each test scenario are weighted and summed to obtain the comprehensive economic score of the scheduling strategy.

5. The new energy grid dispatching strategy economy optimization method according to claim 4, characterized in that, Based on the comprehensive economic evaluation score, select a subset of scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios, including: Select scheduling strategies with an overall economic score lower than a preset threshold, identify the k worst single-scenario economic scores of the selected strategy in each scenario in the test scenario library, and mark them as failed scenarios. Extract feature vectors for failure scenarios, including: new energy output volatility, net load ramp-up rate, and new energy penetration rate. 6.The method of claim 1, wherein, Using a comprehensive economic score and eigenvectors, a directed evolutionary operation is performed on the current policy population to generate a new generation of policy populations, including: The current strategy population is ranked according to the comprehensive economic score, and the M strategies with the highest scores are retained as elite strategies. By performing crossover and mutation operations on elite strategies, a new generation of strategy populations can be generated; Specifically, for the new strategy generated by the mutation operation, its neighboring domain in the feature vector space of the failure scenario is detected; if the failure scenario exists in the neighboring domain, the parameters of the new strategy are adjusted in a targeted manner, and the targeted adjustment includes at least one of the following: increasing the minimum output coefficient of thermal power, increasing the energy storage discharge threshold, decreasing the energy storage charging threshold, and decreasing the tolerance for wind and solar curtailment. 7.The method of claim 4, wherein, The strategy in the iteration is subjected to periodic stress tests using a set of extreme scenarios, and the weights of the extreme scenarios in subsequent optimizations are dynamically adjusted based on the stress test results, including: After each Z iterations of optimization, the strategy with the best overall economic score in the current strategy population will be specifically tested in the set of extreme scenarios. If the economic performance score of the optimal strategy in a single scenario is lower than the preset threshold in a certain extreme scenario, the proportion of that extreme scenario will be increased in the next special test. 8.The method of claim 1, wherein, Based on historical source load data and preset filtering rules, an extreme scenario set is constructed, which includes at least one of the following scenarios: Scenarios where the daily output of new energy sources is less than 10% of the installed capacity, scenarios where the hourly output fluctuation rate of new energy sources exceeds 50%, and peak load scenarios. 9.The method of claim 4, wherein, The iteration stops when any of the following preset termination conditions are met: The preset maximum number of iterations has been reached; Alternatively, the overall economic score of the current strategy population converges in the test scenario library, and in the most recent phased stress test, the single-scenario economic score of the optimal strategy in each extreme scenario is greater than the preset threshold.

10. A new energy grid dispatching strategy economy optimization system, characterized in that, include: The scenario construction module is used to acquire historical source load data, generate an initial scenario set, and cluster and compress it to obtain a test scenario library. In addition, an extreme scenario set is constructed based on historical source load data and preset filtering rules; The initial policy generation module is used to generate multiple scheduling policies, forming an initial policy population; The iterative evolution module is used to perform the following iterative evolution process: Simulate each scheduling strategy in the current strategy population in the test scenario library and calculate the comprehensive economic score of each scheduling strategy; Based on the comprehensive economic score, select some scheduling strategies, identify the failure scenarios corresponding to the selected strategies, and extract the feature vectors corresponding to the failure scenarios. By using a comprehensive economic score and feature vectors, a directed evolutionary operation is performed on the current strategy population to generate a new generation of strategy population; The stress test module is used to conduct phased stress tests on the strategy during the iterative evolution process using a set of extreme scenarios, and dynamically adjust the weight of extreme scenarios in subsequent optimizations based on the stress test results. The strategy output module is used to output the final scheduling strategy when the preset termination conditions are met.

Images