A wind-solar-thermal storage combined system closed-loop optimization scheduling method and system based on post-evaluation feedback

Abstract

The application discloses a wind-solar-thermal-storage combined system closed-loop optimization scheduling method and system based on post-evaluation feedback, and belongs to the technical field of power system automation and new energy optimization scheduling. The device comprises the following steps: based on the multi-source data of the wind-solar-thermal-storage combined system, an optimal output strategy is generated by using a joint optimization module; actual operation data and actual operation results are collected by using a data collection module; a post-evaluation result is generated based on the actual operation data and the actual operation results by using a post-evaluation module; based on the post-evaluation result, an optimization scheduling model is updated by using a feedback control module, and the optimal output strategy of the next scheduling period is generated by using the updated optimization scheduling model, so that the wind-solar-thermal-storage combined system closed-loop optimization scheduling is realized. The application quantitatively compares the effects of the joint operation and a plurality of virtual operation scenes by using the post-evaluation module, and long-term dynamic adjustment of the combined system is realized to avoid model rigidity.

Description

Technical Field

[0001] This invention relates to the field of power system automation and new energy optimization dispatching technology, specifically to a closed-loop optimization dispatching method and system for a wind-solar-thermal-storage integrated system based on post-evaluation feedback. Background Technology

[0002] The intermittent and random nature of wind and solar power output poses a significant challenge to the safe and stable operation of the power grid. To mitigate fluctuations and improve grid absorption capacity, integrated wind, solar, thermal, and energy storage systems have become the mainstream solution.

[0003] Most existing optimization scheduling methods are open-loop or semi-open-loop. The open-loop model is based on predicted data and uses a single optimization model for one-time optimization calculation to obtain and execute the scheduling plan. Once the optimization model is fixed, its parameters and structure are no longer dynamically updated with external operating conditions. After long-term operation, it is easy to deviate from the actual data characteristics, and the deviation will gradually accumulate over time, eventually leading to a significant decrease in prediction or control accuracy and generating large errors. The "semi-open-loop" model performs rolling optimization based on ultra-short-term predictions, but the optimization model, core parameters and algorithms used are pre-set and remain unchanged. They cannot be dynamically adjusted according to actual operating conditions and changes in data distribution. Such models cannot fundamentally determine whether the currently used model and parameters can adapt to the scheduling optimization of the joint system in the long term. They do not have the ability to update the core model. As time goes by and the scenario changes, the applicability of the model gradually decreases, and the model is prone to rigidity, resulting in the accumulation of errors. Summary of the Invention

[0004] This invention proposes a closed-loop optimization scheduling method and system for a wind-solar-thermal-storage integrated system based on post-evaluation feedback. By introducing a multi-dimensional post-evaluation module, the method quantitatively compares the effects of joint operation with various virtual operation scenarios. The evaluation results are then transformed into dynamic adjustment instructions for model algorithms or parameters through a multi-strategy intelligent controller, thereby realizing closed-loop optimization scheduling of the wind-solar-thermal-storage integrated system and improving the long-term economic efficiency of the system.

[0005] To achieve the above objectives, the present invention is implemented using the following technical solution.

[0006] On the one hand, this invention provides a closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system, comprising:

[0007] Based on the historical data of the acquired wind-solar-thermal-storage integrated system, an optimized scheduling model is constructed;

[0008] Based on the multi-source data of the wind-solar-thermal-storage combined system collected at the current moment, the optimal power output strategy of the wind-solar-thermal-storage combined system for the current scheduling cycle is obtained by using the optimization scheduling model for optimization calculation. Based on the optimization scheduling model and historical data, a multi-scenario virtual operation model is constructed.

[0009] In response to the wind-solar-thermal-storage integrated system producing based on the optimal output strategy of the current scheduling cycle, the actual operating data and actual operating results of the current scheduling cycle are collected.

[0010] Based on the actual operation results, predictive operation data is generated using a multi-scenario virtual operation model; and based on the actual operation data and the predicted operation data, calculations are performed using the parameter relationships of the obtained benefit quantification dimension to generate the post-evaluation results for the current scheduling cycle.

[0011] Based on the post-evaluation results, the scheduling model is updated and optimized using a multi-strategy fusion method. The updated optimized scheduling model is then used to generate the optimal power output strategy for the next scheduling cycle based on the multi-source data of the wind-solar-thermal-storage integrated system collected in real time, so as to achieve closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system.

[0012] Optionally, the historical data of the wind-solar-thermal-storage combined system includes system production cost, real-time electricity price, tie-line switching power, curtailed wind power, curtailed solar power, energy storage charging and discharging power, total carbon emissions, spinning reserve capacity, output change, and depth of discharge.

[0013] Optionally, the optimized scheduling model includes an objective function for multi-objective weighted summation of economic, environmental, safety, and equipment life objectives;

[0014] An objective function is constructed based on a preset objective function constraint domain using historical data from a combined wind, solar, thermal, and energy storage system. The calculation expression of the objective function is as follows:

[0015] ;

[0016] in, For economic purposes; Weighting coefficients for economic objectives; For environmental protection goals; Weighting coefficients for environmental protection objectives; For security purposes; Weighting coefficients for security objectives; For equipment lifespan targets; Weighting coefficients for equipment lifespan targets; min represents minimizing the computation, and F is the objective function;

[0017] ;

[0018] ;

[0019] ;

[0020] ;

[0021] Where T is the total number of scheduling periods, and t is the scheduling period; The coal consumption cost of thermal power units during dispatch period t; The real-time electricity price for the dispatch period t; The switching power of the tie-line during scheduling period t; This is the wind curtailment penalty coefficient; The wind curtailment power during the t-schedule period; This is the penalty coefficient for discarded light; The power of abandoned optical signals during scheduling period t; This refers to the cost coefficient for energy storage charging and discharging losses. Let t be the absolute value of the energy storage charging and discharging power during the scheduling period t; The unit carbon emission cost coefficient; The total carbon emissions during scheduling period t; For insufficient rotational reserve penalty coefficient; The required spinning reserve capacity of the system during the t-schedule period; The actual spinning reserve capacity of the system during the t-schedule period; Adjusting the wear coefficient for thermal power units; For the time period t, the change in the output of the thermal power unit is . This is the energy storage lifespan loss coefficient; The depth of energy storage discharge during time period t.

[0022] Optionally, the optimized scheduling model further includes a knowledge base, which stores the operation of updating the optimized scheduling model based on the post-evaluation results using a multi-strategy fusion method.

[0023] When the optimization scheduling model is used to perform optimization calculations based on multi-source data of the current-time wind-solar-thermal-storage integrated system, the parameters of the optimization scheduling model are dynamically adjusted through the knowledge base, including:

[0024] Based on the optimized scheduling model update operation, scene clustering is performed using the maximum likelihood estimation method and the expectation-maximization algorithm to obtain the basic scene data;

[0025] Based on the aforementioned scenario data, predictions are made using an LSTM network to obtain the parameter change trend.

[0026] The overall benefit improvement rate is calculated based on the parameter change trend and the optimized scheduling model update operation.

[0027] Based on the aforementioned comprehensive benefit improvement rate, a dynamic adjustment strategy for the optimized scheduling model is generated using the knowledge distillation method.

[0028] The parameters of the optimized scheduling model are dynamically adjusted using a dynamic adjustment strategy.

[0029] Optionally, the actual operating results include actual wind and solar power output and actual load data;

[0030] Based on the actual operation results, predictive operation data is generated using a multi-scenario virtual operation model, including:

[0031] Based on the actual operating data, virtual independent operating data is calculated using the independent operating data of wind, solar, thermal, and energy storage from historical data.

[0032] Based on the actual operating data, virtual joint operation data is obtained by optimizing the scheduling model.

[0033] Based on the actual operating data, virtual ideal operating data is obtained by offline optimization calculation using a global optimal algorithm.

[0034] Optionally, the post-evaluation results include benefit quantification indicators and correlation evaluation results;

[0035] The dimensions for quantifying benefits include: economic benefits, new energy consumption, model accuracy, energy storage health, environmental benefits, and comprehensive benefits.

[0036] The economic benefit dimension includes the absolute value index of synergistic efficiency, the relative value index of synergistic efficiency, and the optimization potential index.

[0037] The new energy consumption includes consumption increase indicators and consumption rate indicators;

[0038] The model accuracy includes simulation deviation rate and decision consistency index;

[0039] The energy storage health includes cycle life loss indicators and utilization efficiency indicators;

[0040] The environmental benefits include carbon emission reduction indicators and carbon emission intensity indicators;

[0041] The benefit quantification indicators are calculated from the actual operating data and the predicted operating data using the parameter relationships of the benefit quantification dimensions, including:

[0042] The absolute value index of synergistic efficiency is calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0043] ;

[0044] in, This represents the total cost of virtual independent operations in the virtual independent operation data. This refers to the actual total operating cost in the actual operating data;

[0045] Based on the absolute value index of synergistic efficiency and the virtual independent operation data, calculate the relative value index of synergistic efficiency. The calculation formula is:

[0046] ;

[0047] Calculate the optimization potential index based on the virtual ideal operating data and the actual operating data. The calculation formula is:

[0048] ;

[0049] in, The total cost of running a virtual ideal system based on virtual ideal system operating data;

[0050] Calculate the absorption capacity improvement index based on the virtual independent operation data and actual operation data. The calculation formula is:

[0051] ;

[0052] in, This refers to the total amount of energy wasted by virtual independent operations in the virtual independent operation data; This refers to the total amount of energy wasted during actual operation, as shown in the actual operational data.

[0053] The absorption rate index is calculated based on the actual operating data. The calculation formula is:

[0054] ;

[0055] in, This refers to the total amount of available new energy sources in the actual operational data;

[0056] The simulation deviation rate index is calculated based on virtual joint operation data and actual operation data. The calculation formula is:

[0057] ;

[0058] in, The total cost calculated for the virtual joint operation model;

[0059] Calculate the decision consistency index based on the virtual joint operation data and the actual operation data. The calculation formula is:

[0060] ;

[0061] in, For virtual joint operation contribution plans in virtual joint operation data; To contribute to the actual execution of the actual operating data;

[0062] Calculate cycle life loss index based on the actual operating data. The calculation formula is:

[0063] ;

[0064] in, This is the loss coefficient; For exponential coefficients; The discharge depth of the i-th cycle in the actual operating data;

[0065] The utilization efficiency index is calculated based on the actual operating data and the virtual independent operating data. The calculation formula is as follows:

[0066] ;

[0067] in, This refers to the actual total discharge in the actual operating data; The rated energy storage capacity in the virtual independent operating data;

[0068] Carbon emission reduction targets are calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0069] ;

[0070] in, This refers to the carbon emissions of virtual independent operations (VIOs) within the VIO data. This refers to the actual carbon emissions from actual operation in the operational data.

[0071] Carbon emission intensity index calculated based on actual operational data The calculation formula is:

[0072] ;

[0073] in, This refers to the actual total discharge in the actual operating data;

[0074] The comprehensive benefit index is calculated based on the aforementioned synergistic efficiency relative value index, absorption rate index, simulation deviation rate index, cycle life loss index, and carbon emission intensity index to determine the comprehensive benefit dimension. The calculation formula is:

[0075] ;

[0076] in, This is the economic benefit weighting coefficient; This is the weighting coefficient for absorption benefits; These are the model accuracy weighting coefficients; This is the weighting coefficient for energy storage lifetime; Environmental benefit weighting coefficient; This is the preset baseline carbon emission intensity.

[0077] Optionally, based on the aforementioned benefit quantification indicators, correlation evaluation results are generated through trend analysis and anomaly detection methods, including:

[0078] Based on the aforementioned benefit quantification indicators, the data characteristics of the benefit quantification indicators are calculated using a sliding window mechanism;

[0079] The operational trend of the decision-making system is based on the data characteristics of the aforementioned benefit quantification indicators;

[0080] Based on the aforementioned data characteristics, anomalies in the quantification of benefits are identified using a cumulative sum algorithm.

[0081] Based on the system's operational trends and abnormal changes, correlation evaluation results are generated by calculating the correlation coefficient between the benefit quantification indicators and external factors.

[0082] Optionally, the multi-strategy fusion method includes a rule base parameter fine-tuning strategy, a fuzzy logic multi-parameter collaborative adjustment strategy, a reinforcement learning self-optimization strategy, and a Bayesian optimization hyperparameter tuning strategy.

[0083] Optionally, based on the post-evaluation results, the scheduling model is updated and optimized using a multi-strategy fusion method, including:

[0084] Based on the aforementioned benefit quantification indicators and preset error thresholds, a feedback adjustment method for optimizing the scheduling model is determined through a hierarchical triggering mechanism. The feedback adjustment method includes a model adjustment method and a parameter adjustment method.

[0085] The model adjustment method updates and optimizes the scheduling model algorithm based on the post-evaluation results using a multi-strategy fusion method; the parameter adjustment method updates and optimizes the scheduling model parameters based on the post-evaluation results using a multi-strategy fusion method.

[0086] In a second aspect, the present invention provides a closed-loop optimization scheduling system for a wind-solar-thermal-storage integrated system based on post-evaluation feedback, including a storage medium and a processor;

[0087] The storage medium is used to store instructions;

[0088] The processor is configured to operate according to the instructions to execute the closed-loop optimization scheduling method for the wind-solar-thermal-storage integrated system as described in any one of the first aspects above.

[0089] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0090] This invention accurately quantifies the economic benefits of joint operation by comparing actual operational data with predicted operational data generated using a multi-scenario virtual operation model. It achieves precise assessment of optimization potential and identifies model accuracy defects, making collaborative value calculable, model problems locatable, and optimization directions clear. By updating the optimized scheduling model through a multi-strategy fusion method, the model algorithm or parameters can be updated based on post-evaluation results, improving the long-term adaptive capability of the optimized scheduling model for the wind-solar-thermal-storage integrated system. In each scheduling cycle, this invention updates the optimized scheduling model through a multi-strategy fusion method and uses the updated optimized scheduling model to generate the optimal output strategy for the next scheduling cycle based on real-time collected multi-source data from the wind-solar-thermal-storage integrated system. This achieves closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system and ensures dynamic updates of the optimized scheduling model.

[0091] This invention improves the dynamic adjustment and continuous updating capabilities of the optimized scheduling model by storing the optimization scheduling model's update operations in a knowledge base, after each scheduling cycle. It achieves joint optimization of economic, environmental, safety, and equipment lifespan objectives through the objective function of the optimized scheduling model. Furthermore, this invention utilizes trend analysis and anomaly detection methods to calculate and evaluate benefit quantification indicators, clarifying the model optimization objectives and improving the accuracy of updating the optimized scheduling model using multi-strategy fusion methods. Attached Figure Description

[0092] Figure 1 This diagram illustrates the overall architecture of a closed-loop optimization scheduling system. Detailed Implementation

[0093] 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.

[0094] The term "and / or" is merely a description of the parametric relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0095] Example 1

[0096] This embodiment provides a closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system, specifically including:

[0097] Based on the historical data of the acquired wind-solar-thermal-storage integrated system, an optimized scheduling model is constructed;

[0098] Step S1: Based on the multi-source data of the wind-solar-thermal-storage combined system collected at the current moment, generate an optimal output plan for each unit within a scheduling cycle using the current version of the optimized scheduling model, that is, the optimal output strategy of the wind-solar-thermal-storage combined system for the current scheduling cycle.

[0099] Step S2: Construct a multi-scenario virtual operation model based on the current version of the optimized scheduling model and historical data; the multi-scenario virtual operation model is updated in each scheduling cycle according to the updated optimized scheduling model.

[0100] Step S3: The wind-solar-thermal-storage integrated system produces according to the optimal output strategy of the current scheduling cycle, and collects the actual operating data and actual operating results of the current scheduling cycle during the production process of the wind-solar-thermal-storage integrated system. The actual operating results include actual wind and solar power output and actual load data. The actual operating data includes actual thermal power generation and coal consumption, actual energy storage charging and discharging power and SOC, actual total operating cost of the system, wind and solar curtailment, carbon emissions, and equipment operating status parameters.

[0101] Step S4: Based on the actual operation results, generate predicted operation data for various scenarios using a multi-scenario virtual operation model;

[0102] Step S5: Based on the actual operating data and the predicted operating data under multiple scenarios, the post-evaluation results of the current scheduling cycle are generated by using the parameter relationships of the obtained benefit quantification dimension. By comparing the actual operating data and the predicted operating data generated by the multi-scenario virtual operating model, the economic benefits brought by joint operation are accurately quantified, the space for optimization and improvement is accurately assessed, and the model accuracy defects can be identified.

[0103] Step S6: Based on the post-evaluation results, update and optimize the scheduling model using a multi-strategy fusion method.

[0104] Repeat steps S1 to S6 to achieve closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system.

[0105] In summary, in this embodiment, the scheduling model is updated and optimized in each scheduling cycle through a multi-strategy fusion method. The updated optimized scheduling model is then used to generate the optimal output strategy for the next scheduling cycle based on the multi-source data of the wind-solar-thermal-storage integrated system collected in real time. This achieves closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system and ensures the dynamic updating of the optimized scheduling model.

[0106] Example 2

[0107] This embodiment provides a closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system, specifically including:

[0108] Based on the historical data acquired from the wind-solar-thermal-storage integrated system, an optimized scheduling model is constructed. The historical data includes system production costs, real-time electricity prices, tie-line switching power, curtailed wind power, curtailed solar power, energy storage charging and discharging power, total carbon emissions, spinning reserve capacity, power output variation, and depth of discharge. The optimized scheduling model includes an objective function and a knowledge base.

[0109] The objective function is used to perform a weighted summation of multiple objectives, including economic, environmental, safety, and equipment life objectives. The specific expression is as follows:

[0110] ;

[0111] in, For economic purposes; Weighting coefficients for economic objectives; For environmental protection goals; Weighting coefficients for environmental protection objectives; For security purposes; Weighting coefficients for security objectives; For equipment lifespan targets; Weighting coefficients for equipment lifespan targets; min represents minimizing the computation, and F is the objective function;

[0112] ;

[0113] ;

[0114] ;

[0115] ;

[0116] Where T is the total number of scheduling periods, and t is the scheduling period; The coal consumption cost of thermal power units during dispatch period t; The real-time electricity price for the dispatch period t; The switching power of the tie-line during scheduling period t; This is the wind curtailment penalty coefficient; The wind curtailment power during the t-schedule period; This is the penalty coefficient for discarded light; The power of abandoned optical signals during scheduling period t; This refers to the cost coefficient for energy storage charging and discharging losses. Let t be the absolute value of the energy storage charging and discharging power during the scheduling period t; The unit carbon emission cost coefficient; The total carbon emissions during scheduling period t; For insufficient rotational reserve penalty coefficient; The required spinning reserve capacity of the system during the t-schedule period; The actual spinning reserve capacity of the system during the t-schedule period; Adjusting the wear coefficient for thermal power units; For the time period t, the change in the output of the thermal power unit is . This is the energy storage lifespan loss coefficient; The depth of energy storage discharge during time period t.

[0117] The preset objective function constraint domain includes: power balance constraint, thermal power unit constraint, wind power unit constraint, photovoltaic unit constraint, energy storage system constraint, tie line power constraint, and spinning reserve constraint.

[0118] Power balance constraints:

[0119] ;

[0120] in, The actual wind power output during time period t; The actual photovoltaic output during period t; For thermal power output during period t; The energy storage discharge power during time period t; The switching power of the tie line during time period t; The load power during time period t; The energy storage charging power is t.

[0121] Thermal power unit constraints:

[0122] ;

[0123] ;

[0124] in, Minimize the technical output of thermal power units; This is the maximum output of the thermal power unit; This represents the maximum ramp rate of the thermal power unit.

[0125] Constraints of energy storage systems:

[0126] ;

[0127] ;

[0128] in, This represents the maximum charging and discharging power of the energy storage. The energy storage state of charge during time period t; This is the limit of the energy storage state of charge; This represents the upper limit of the energy storage state of charge.

[0129] ;

[0130] ;

[0131] in, Forecast wind power output for period t; The photovoltaic power output is predicted for period t.

[0132] The economic targets include coal consumption costs for thermal power plants, switching power costs for tie lines, unit start-up and shutdown costs, wind and solar curtailment penalty costs, and energy storage loss costs; the environmental targets include the environmental costs of various emissions; the safety targets include insufficient spinning reserve penalties, voltage deviation penalties, and frequency deviation penalties; and the equipment life targets include thermal power unit regulation wear costs and energy storage discharge depth loss costs.

[0133] The knowledge base is used to store the scheduling model operation updated and optimized based on the post-evaluation results using a multi-strategy fusion method.

[0134] When the optimization scheduling model is used to perform optimization calculations based on multi-source data of the current-time wind-solar-thermal-storage integrated system, the parameters of the optimization scheduling model are dynamically adjusted through the knowledge base, including:

[0135] Based on the optimized scheduling model update operation, scene clustering is performed using the maximum likelihood estimation method and the expectation-maximization algorithm to obtain the basic scene data;

[0136] Based on the aforementioned scenario data, predictions are made using an LSTM network to obtain the parameter change trend.

[0137] The overall benefit improvement rate is calculated based on the parameter change trend and the optimized scheduling model update operation.

[0138] Based on the aforementioned comprehensive benefit improvement rate, a dynamic adjustment strategy for the optimized scheduling model is generated using the knowledge distillation method.

[0139] The parameters of the optimized scheduling model are dynamically adjusted using a dynamic adjustment strategy.

[0140] The knowledge base's specific structure includes a typical scenario-optimal parameter mapping library, a parameter evolution trajectory library, an adjustment strategy effect evaluation library, and a knowledge distillation library.

[0141] Typical scenarios-optimal parameter mapping library, using Gaussian mixture model for scene clustering, and weighted Euclidean distance to calculate scene similarity;

[0142] A parameter evolution trajectory library is used to predict the future trend of parameter changes using an LSTM network.

[0143] Adjust the strategy effectiveness evaluation library to evaluate the effectiveness of each strategy adjustment and calculate the overall benefit improvement rate after the adjustment;

[0144] Knowledge distillation libraries compress the policy knowledge learned by deep reinforcement learning models into lightweight rules or shallow neural networks using knowledge distillation techniques.

[0145] Based on the multi-source data of the wind-solar-thermal-storage combined system collected at the current moment, the optimal power output strategy of the wind-solar-thermal-storage combined system for the current scheduling cycle is obtained by using the optimization scheduling model for optimization calculation. Based on the optimization scheduling model and historical data, a multi-scenario virtual operation model is constructed.

[0146] Production is carried out using the combined wind, solar, thermal, and energy storage system based on the optimal output strategy for the current scheduling cycle, and the actual operating data and results for the current scheduling cycle are collected.

[0147] Based on the actual operation results, predictive operation data is generated using a multi-scenario virtual operation model.

[0148] Based on the actual and predicted operating data, calculations are performed using the parameter relationships of the acquired benefit quantification dimensions to generate the post-evaluation results for the current scheduling cycle. The benefit quantification dimensions include: economic benefit dimension, new energy consumption dimension, model accuracy dimension, energy storage health dimension, environmental benefit dimension, and comprehensive benefit dimension.

[0149] The economic benefit dimension includes the absolute value index of synergistic efficiency, the relative value index of synergistic efficiency, and the optimization potential index.

[0150] The new energy consumption includes consumption increase indicators and consumption rate indicators;

[0151] The model accuracy includes simulation deviation rate and decision consistency index;

[0152] The energy storage health includes cycle life loss indicators and utilization efficiency indicators;

[0153] The environmental benefits include carbon emission reduction indicators and carbon emission intensity indicators.

[0154] The post-evaluation results include quantitative indicators of benefits and correlation evaluation results.

[0155] The benefit quantification indicators are calculated from the actual operating data and the predicted operating data using the parameter relationships of the benefit quantification dimensions, including:

[0156] The absolute value index of synergistic efficiency is calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0157] ;

[0158] in, This represents the total cost of virtual independent operations in the virtual independent operation data. This refers to the actual total operating cost in the actual operating data;

[0159] Based on the absolute value index of synergistic efficiency and the virtual independent operation data, calculate the relative value index of synergistic efficiency. The calculation formula is:

[0160] ;

[0161] Calculate the optimization potential index based on the virtual ideal operating data and the actual operating data. The calculation formula is:

[0162] ;

[0163] in, The total cost of running a virtual ideal system based on virtual ideal system operating data;

[0164] Calculate the absorption capacity improvement index based on the virtual independent operation data and actual operation data. The calculation formula is:

[0165] ;

[0166] in, This refers to the total amount of energy wasted by virtual independent operations in the virtual independent operation data; This refers to the total amount of energy wasted during actual operation, as shown in the actual operational data.

[0167] The absorption rate index is calculated based on the actual operating data. The calculation formula is:

[0168] ;

[0169] in, This refers to the total amount of available new energy sources in the actual operational data;

[0170] The simulation deviation rate index is calculated based on virtual joint operation data and actual operation data. The calculation formula is:

[0171] ;

[0172] in, The total cost calculated for the virtual joint operation model;

[0173] Calculate the decision consistency index based on the virtual joint operation data and the actual operation data. The calculation formula is:

[0174] ;

[0175] in, For virtual joint operation contribution plans in virtual joint operation data; To contribute to the actual execution of the actual operating data;

[0176] Calculate cycle life loss index based on the actual operating data. The calculation formula is:

[0177] ;

[0178] in, This is the loss coefficient; For exponential coefficients; The discharge depth of the i-th cycle in the actual operating data;

[0179] The utilization efficiency index is calculated based on the actual operating data and the virtual independent operating data. The calculation formula is as follows:

[0180] ;

[0181] in, This refers to the actual total discharge in the actual operating data; The rated energy storage capacity in the virtual independent operating data;

[0182] Carbon emission reduction targets are calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0183] ;

[0184] in, This refers to the carbon emissions of virtual independent operations (VIOs) within the VIO data. This refers to the actual carbon emissions from actual operation in the operational data.

[0185] Carbon emission intensity index calculated based on actual operational data The calculation formula is:

[0186] ;

[0187] in, This refers to the actual total discharge in the actual operating data;

[0188] The comprehensive benefit index is calculated based on the aforementioned synergistic efficiency relative value index, absorption rate index, simulation deviation rate index, cycle life loss index, and carbon emission intensity index to determine the comprehensive benefit dimension. The calculation formula is:

[0189] ;

[0190] in, This is the economic benefit weighting coefficient; This is the weighting coefficient for absorption benefits; These are the model accuracy weighting coefficients; This is the weighting coefficient for energy storage lifetime; Environmental benefit weighting coefficient; This is the preset baseline carbon emission intensity.

[0191] The correlation evaluation results are generated based on the benefit quantification indicators through trend analysis and anomaly detection methods, including:

[0192] Based on the aforementioned benefit quantification indicators, the data characteristics of the benefit quantification indicators are calculated using a sliding window mechanism;

[0193] The operational trend of the decision-making system is based on the data characteristics of the aforementioned benefit quantification indicators;

[0194] Based on the aforementioned data characteristics, anomalies in the quantification of benefits are identified using a cumulative sum algorithm.

[0195] Based on the system's operational trends and abnormal changes, correlation evaluation results are generated by calculating the correlation coefficient between the benefit quantification indicators and external factors.

[0196] Based on the post-evaluation results, the scheduling model is updated and optimized through a multi-strategy fusion method, specifically: the multi-strategy fusion method includes a rule base parameter fine-tuning strategy, a fuzzy logic multi-parameter collaborative adjustment strategy, a reinforcement learning self-optimization strategy, and a Bayesian optimization hyperparameter tuning strategy.

[0197] Based on the post-evaluation results, the scheduling model is updated and optimized using a multi-strategy fusion method, including:

[0198] Based on the aforementioned benefit quantification indicators and preset error thresholds, a feedback adjustment method for optimizing the scheduling model is determined through a hierarchical triggering mechanism. The feedback adjustment method includes a model adjustment method and a parameter adjustment method.

[0199] The model adjustment method updates and optimizes the scheduling model algorithm based on the post-evaluation results using a multi-strategy fusion method.

[0200] The parameter adjustment method updates and optimizes the scheduling model parameters based on the post-evaluation results using a multi-strategy fusion method.

[0201] The parameter fine-tuning strategy based on the rule base uses IF-THEN rules for storage, maps post-evaluation metrics to parameter adjustment actions, and supports online updates of rule confidence.

[0202] A multi-parameter collaborative adjustment strategy based on fuzzy logic is adopted, and a multi-input multi-output fuzzy control system is used to obtain precise adjustment amounts through fuzzification, fuzzy inference and defuzzification.

[0203] Based on reinforcement learning, a self-optimizing strategy is adopted, using a deep deterministic policy gradient algorithm. The state space is a post-evaluation index vector, the action space is a model parameter adjustment vector, and the reward function is a function of the comprehensive benefit index.

[0204] The hyperparameter tuning strategy based on Bayesian optimization uses Gaussian process regression as a surrogate model and expected value enhancement as the acquisition function to periodically tune key hyperparameters.

[0205] The updated optimized scheduling model generates the optimal power output strategy for the next scheduling cycle based on multi-source data collected in real time from the wind-solar-thermal-storage integrated system, thereby achieving closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system.

[0206] Example 3

[0207] Based on the same inventive concept as Embodiment 1, this embodiment introduces a closed-loop optimization scheduling system for a wind-solar-thermal-storage integrated system based on post-evaluation feedback, including a storage medium and a processor, such as... Figure 1 As shown, the processor includes:

[0208] The joint optimization module is used to perform optimization calculations based on the multi-source data of the wind-solar-thermal-storage combined system collected at the current moment, and to obtain the optimal power output strategy of the wind-solar-thermal-storage combined system for the current scheduling cycle.

[0209] The data acquisition module is used to acquire multi-source data from the wind-solar-thermal-storage integrated system in real time, and to collect the actual operating data and actual operating results of the current scheduling cycle.

[0210] The post-evaluation module is used to generate predicted operation data based on the actual operation results using a multi-scenario virtual operation model; and to calculate the post-evaluation results for the current scheduling cycle based on the actual operation data and the predicted operation data using the parameter relationships of the obtained benefit quantification dimension.

[0211] The feedback control module is used to update and optimize the scheduling model based on the post-evaluation results using a multi-strategy fusion method.

[0212] The post-evaluation module implements parallel simulation calculation based on a distributed computing framework; and performs multi-dimensional post-evaluation analysis based on actual operating data and results, including multi-scenario virtual operation simulation, multi-dimensional benefit quantification index calculation, trend analysis and anomaly detection.

[0213] The multi-scenario virtual operation simulation includes:

[0214] Scenario A: Virtual independent operation, assuming that wind, solar, thermal and energy storage operate independently without coordination, and energy storage does not participate in regulation;

[0215] Scenario B: Virtual joint operation, using the same model and parameters as the optimization scheduling model in the joint optimization module, but with actual data as input for further optimization;

[0216] Scenario C: Virtual ideal operation, assuming perfect prediction, offline optimization using a global optimization algorithm to obtain the theoretical upper limit of the operating effect.

[0217] The multi-dimensional benefit quantification indicators include: economic benefit indicators, new energy consumption indicators, model accuracy indicators, energy storage health indicators, environmental benefit indicators, and comprehensive benefit index. The economic benefit indicators include the absolute value of synergistic efficiency, the relative value of synergistic efficiency, and the optimization potential index. The new energy consumption indicators include the increase in consumption and the consumption rate. The model accuracy indicators include simulation deviation rate and decision consistency. The energy storage health indicators include cycle life loss and utilization efficiency. The environmental benefit indicators include carbon emission reduction and carbon emission intensity. The trend analysis and anomaly detection include: sliding window statistical analysis, CUSUM mutation point detection, periodic decomposition, and correlation analysis. The scenario clustering and pattern recognition use a Gaussian mixture model or K-means algorithm to cluster historical scheduling cycles according to feature vectors, establishing a typical operation scenario library.

[0218] The feedback control module generates adjustment instructions for the optimized scheduling model or parameters based on the post-evaluation results through multi-strategy fusion; the feedback control module adopts a hierarchical triggering mechanism, including:

[0219] The first level of routine adjustment involves fine-tuning parameters when indicators fluctuate within the normal range.

[0220] The second layer of abnormal adjustment involves refactoring parameters when indicators exceed the warning threshold.

[0221] The third major adjustment is to update the model when indicators continue to deteriorate or when structural changes occur in the system.

[0222] When executing the multi-strategy fusion method, one or more combinations of the following strategies are executed:

[0223] The parameter fine-tuning strategy based on the rule base uses IF-THEN rules for storage, maps post-evaluation metrics to parameter adjustment actions, and supports online updates of rule confidence.

[0224] A multi-parameter collaborative adjustment strategy based on fuzzy logic is adopted, and a multi-input multi-output fuzzy control system is used to obtain precise adjustment amounts through fuzzification, fuzzy inference and defuzzification.

[0225] Based on reinforcement learning, a self-optimizing strategy is adopted, using a deep deterministic policy gradient algorithm. The state space is a post-evaluation index vector, the action space is a model parameter adjustment vector, and the reward function is a function of the comprehensive benefit index.

[0226] The hyperparameter tuning strategy based on Bayesian optimization uses Gaussian process regression as a surrogate model and expected value enhancement as the acquisition function to periodically tune key hyperparameters.

[0227] The data acquisition module interfaces with the SCADA system and the power acquisition system.

[0228] The joint optimization module uses the current version of the optimized scheduling model based on data such as wind and solar power output forecasts, load forecasts, market electricity prices, and equipment status. and its parameter set With the goal of minimizing the total system operating cost and the curtailment rate of wind and solar power, optimization calculations are performed to obtain the optimal output plan for each unit of wind, solar, thermal, and energy storage in the next scheduling cycle.

[0229] Optimize scheduling model The parameter set includes the objective function and constraints. This includes various weighting coefficients, penalty factors, and equipment operating boundary parameters.

[0230] The data acquisition module system executes the power output plan and collects actual operating data after a scheduling cycle, including: actual wind and solar power output, actual load, actual thermal power generation and coal consumption, actual energy storage charging and discharging power and SOC, actual total operating cost of the system, amount of wind and solar curtailment, carbon emissions, equipment operating status parameters, etc.

[0231] The post-evaluation module is used to perform the following operations:

[0232] (1) Multi-scenario virtual operation simulation

[0233] Using the actual wind and solar power output and actual load data collected in S2 as unchangeable boundary conditions, various operating scenarios are simulated:

[0234] Scenario A: Virtual Independent Operation: Assume wind, solar, thermal, and energy storage operate independently without coordination. Wind and solar power are fully fed into the grid, load is met by thermal power and purchased electricity, and energy storage does not participate in regulation. Calculate the total system cost under this scenario. Wind and solar curtailment Carbon emissions .

[0235] Scenario B Virtual Joint Operation: Using the same model as in S1 and parameters However, if the actual data is used as input and the optimization calculation is performed again, the cost is denoted as... The amount of wind and solar power curtailed is recorded as Carbon emissions are recorded as .

[0236] Scenario C: Virtual Ideal Operation: Assuming perfect predictive capability, offline optimization is performed using a globally optimal algorithm to obtain the theoretical upper limit of the operating performance, which serves as the evaluation benchmark. The cost is denoted as... .

[0237] (2) Multi-dimensional benefit quantification indicator system

[0238] The absolute value index of synergistic efficiency is calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0239] ;

[0240] in, This represents the total cost of virtual independent operations in the virtual independent operation data. This refers to the actual total operating cost in the actual operating data;

[0241] Based on the absolute value index of synergistic efficiency and the virtual independent operation data, calculate the relative value index of synergistic efficiency. The calculation formula is:

[0242] ;

[0243] Calculate the optimization potential index based on the virtual ideal operating data and the actual operating data. The calculation formula is:

[0244] ;

[0245] in, The total cost of running a virtual ideal system based on virtual ideal system operating data;

[0246] Calculate the absorption capacity improvement index based on the virtual independent operation data and actual operation data. The calculation formula is:

[0247] ;

[0248] in, This refers to the total amount of energy wasted by virtual independent operations in the virtual independent operation data; This refers to the total amount of energy wasted during actual operation, as shown in the actual operational data.

[0249] The absorption rate index is calculated based on the actual operating data. The calculation formula is:

[0250] ;

[0251] in, This refers to the total amount of available new energy sources in the actual operational data;

[0252] The simulation deviation rate index is calculated based on virtual joint operation data and actual operation data. The calculation formula is:

[0253] ;

[0254] in, The total cost calculated for the virtual joint operation model;

[0255] Calculate the decision consistency index based on the virtual joint operation data and the actual operation data. The calculation formula is:

[0256] ;

[0257] in, For virtual joint operation contribution plans in virtual joint operation data; To contribute to the actual execution of the actual operating data;

[0258] Calculate cycle life loss index based on the actual operating data. The calculation formula is:

[0259] ;

[0260] in, This is the loss coefficient; For exponential coefficients; The discharge depth of the i-th cycle in the actual operating data;

[0261] The utilization efficiency index is calculated based on the actual operating data and the virtual independent operating data. The calculation formula is as follows:

[0262] ;

[0263] in, This refers to the actual total discharge in the actual operating data; The rated energy storage capacity in the virtual independent operating data;

[0264] Carbon emission reduction targets are calculated based on virtual independent operation data and actual operation data. The calculation formula is:

[0265] ;

[0266] in, This refers to the carbon emissions of virtual independent operations (VIOs) within the VIO data. This refers to the actual carbon emissions from actual operation in the operational data.

[0267] Carbon emission intensity index calculated based on actual operational data The calculation formula is:

[0268] ;

[0269] in, This refers to the actual total discharge in the actual operating data;

[0270] The comprehensive benefit index is calculated based on the aforementioned synergistic efficiency relative value index, absorption rate index, simulation deviation rate index, cycle life loss index, and carbon emission intensity index to determine the comprehensive benefit dimension. The calculation formula is:

[0271] ;

[0272] in, This is the economic benefit weighting coefficient; This is the weighting coefficient for absorption benefits; These are the model accuracy weighting coefficients; This is the weighting coefficient for energy storage lifetime; Environmental benefit weighting coefficient; This is the preset baseline carbon emission intensity.

[0273] (3) Trend analysis and anomaly detection

[0274] 1) Sliding window statistical analysis: Calculate the mean, variance, and rate of change of each indicator over the past N scheduling cycles to determine the stability and trend of system operation.

[0275] 2) Mutation point detection: The cumulative sum algorithm is used to detect abnormal changes in system performance indicators and identify equipment failures or sudden changes in the external environment.

[0276] 3) Periodic decomposition: Decompose the time series of benefit indicators into trend, seasonal and random components to identify long-term change patterns and seasonal characteristics.

[0277] 4) Correlation analysis: Calculate the correlation coefficient between each evaluation indicator and external factors to identify key influencing factors.

[0278] (4) Scene clustering and pattern recognition

[0279] K-means or DBSCAN clustering algorithms are used to cluster historical scheduling cycles according to characteristics such as weather type, load characteristics, and electricity price fluctuations, and a typical operation scenario library is established to provide a basis for subsequent parameter matching.

[0280] The feedback control module, based on the post-evaluation results, generates an optimized scheduling model through multi-strategy fusion. or parameter set The adjustment instructions. If If the value remains negative or below the expected threshold, it indicates that the current joint operation strategy is not even as effective as operating alone, or the efficiency is poor, and the feedback controller is activated.

[0281] Parameter tuning: Automatic model adjustment Weighting coefficients and penalty factors in the model, for example, increasing the cost of wind curtailment penalties, can make the model more inclined to absorb new energy sources.

[0282] Model update: In more severe cases, model reconstruction can be triggered, switching from a single-objective model focused on economic efficiency to a multi-objective model that combines economic efficiency and environmental protection, and introducing machine learning algorithms to replace the original optimization algorithms.

[0283] If the actual operating cost and the virtual joint operating cost deviate too much, it indicates that the model has a poor fit to the actual situation. The feedback controller should adjust the compensation parameters for the prediction error or the conservatism parameters in the robust optimization.

[0284] The updated optimization model and parameter set The joint optimization module is then applied to the next scheduling cycle. This process repeats continuously, forming a closed-loop system that constantly improves itself.

[0285] The scenario is as follows: Taking a wind-solar-thermal-storage combined system in a certain region as an example, the total installed capacity is: 200MW wind power, 150MW photovoltaic power, 300MW thermal power, and 50MW / 200MWh energy storage.

[0286] Initial state: Optimization model For a linear programming model with the objective of minimizing total operating cost, the parameter set... The penalty cost for wind curtailment is 0.3 yuan / kWh, the penalty cost for solar curtailment is 0.25 yuan / kWh, the energy storage loss cost is 0.1 yuan / kWh, and the carbon cost is 0.05 yuan / kg.

[0287] Day 1 Operation:

[0288] The system executes a daily scheduling plan of 96 points, collects data after actual operation, and calculates the actual operating cost. The cost was 985,000 yuan, with a wind curtailment rate of 12%, a solar curtailment rate of 8%, and carbon emissions of 420 tons.

[0289] The post-evaluation module performs simulation calculations: virtual independent operating cost. The cost was 1.182 million yuan, with a wind curtailment rate of 0% (full grid connection), a solar curtailment rate of 0%, carbon emissions of 510 tons, and virtual joint operation costs. The virtual ideal operating cost is 1.013 million yuan. It amounted to 927,000 yuan.

[0290] Calculate each indicator:

[0291] Ten thousand yuan, .

[0292] ;

[0293] ;

[0294] ;

[0295] ;

[0296] Analysis of the feedback control module revealed that, The value is positive but lower than the historical average of 220,000 yuan; the wind curtailment rate of 12% is higher than the target value of 8%; and the optimization potential index of 6.26% indicates that there is still room for improvement. According to the rule base, the wind curtailment penalty cost adjustment is triggered: The price is 0.06 yuan / kWh. Meanwhile, the fuzzy logic controller adjusts the input... , Output , Slightly reduce the weighting of economic factors. Adjusted parameter set. The penalty for curtailing wind power is 0.36 yuan / kWh, the penalty for curtailing solar power is 0.28 yuan / kWh, the economic weight is 0.48 (original weight 0.5), and the environmental weight is 0.12 (original weight 0.1).

[0297] The day's operational data is stored in the knowledge base, and the scene features [summer, sunny, high load, high photovoltaic] are clustered into scene S001, and the parameter statistics information of this scene is updated.

[0298] Day 2 Operation:

[0299] Use the updated parameter set Optimization was performed. After actual operation, The wind curtailment rate was 9.5%, the solar curtailment rate was 6.8%, and the carbon emissions were 405 tons. The comprehensive benefit index increased from 0.86 to 0.89, with an increase of 10,000 yuan.

[0300] After 30 days of continuous iteration:

[0301] System parameters gradually converged: wind curtailment penalty stabilized at 0.41-0.43 yuan / kWh, solar curtailment penalty at 0.32-0.34 yuan / kWh, economic efficiency weight at 0.45-0.47, and environmental protection weight at 0.13-0.14. (Average) The cost was increased to 235,000 yuan, the wind curtailment rate dropped to 7.2%, the solar curtailment rate was 5.1%, and the comprehensive benefit index remained stable at 0.93-0.95.

[0302] In summary, this invention improves the dynamic adjustment capability and continuous update capability of the optimized scheduling model by storing the optimized scheduling model in a knowledge base, thereby enhancing its ability to continuously update after each scheduling cycle. It achieves joint optimization of economic, environmental, safety, and equipment lifespan objectives through the objective function of the optimized scheduling model. Furthermore, it utilizes trend analysis and anomaly detection methods to calculate and evaluate benefit quantification indicators, clarifying the model optimization objectives and improving the accuracy of updating the optimized scheduling model using a multi-strategy fusion method.

[0303] 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.

[0304] 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 1A device that provides the functions specified in one or more boxes.

[0305] 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.

[0306] 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.

[0307] 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 closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system, characterized in that, include: Based on the historical data of the acquired wind-solar-thermal-storage integrated system, an optimized scheduling model is constructed; Based on the multi-source data of the wind-solar-thermal-storage combined system collected at the current moment, the optimal power output strategy of the wind-solar-thermal-storage combined system for the current scheduling cycle is obtained by using an optimized scheduling model for optimization calculation. Based on the optimized scheduling model and historical data, a multi-scenario virtual operation model is constructed. In response to the wind-solar-thermal-storage integrated system producing based on the optimal output strategy of the current scheduling cycle, the actual operating data and actual operating results of the current scheduling cycle are collected. Based on the actual operation results, predictive operation data is generated using a multi-scenario virtual operation model; and based on the actual operation data and the predicted operation data, calculations are performed using the parameter relationships of the obtained benefit quantification dimension to generate the post-evaluation results of the current scheduling cycle. Based on the post-evaluation results, the scheduling model is updated and optimized using a multi-strategy fusion method. The updated optimized scheduling model is then used to generate the optimal power output strategy for the next scheduling cycle based on the multi-source data of the wind-solar-thermal-storage integrated system collected in real time, so as to achieve closed-loop optimized scheduling of the wind-solar-thermal-storage integrated system.

2. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 1, characterized in that, The historical data of the wind-solar-thermal-storage integrated system includes system production cost, real-time electricity price, tie-line switching power, curtailed wind power, curtailed solar power, energy storage charging and discharging power, total carbon emissions, spinning reserve capacity, output change, and depth of discharge.

3. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 2, characterized in that, The optimized scheduling model includes an objective function for multi-objective weighted summation of economic, environmental, safety, and equipment life objectives; An objective function is constructed based on a preset objective function constraint domain using historical data from a combined wind, solar, thermal, and energy storage system. The calculation expression of the objective function is as follows: ； in, For economic purposes; Weighting coefficients for economic objectives; For environmental protection goals; Weighting coefficients for environmental protection objectives; For security purposes; Weighting coefficients for security objectives; For equipment lifespan targets; Weighting coefficients for equipment lifespan targets; min represents the minimization calculation, and F is the objective function; ； ； ； ； Where T is the total number of scheduling periods, and t is the scheduling period; The coal consumption cost of thermal power units during dispatch period t; The real-time electricity price for the dispatch period t; The switching power of the tie-line during scheduling period t; This is the wind curtailment penalty coefficient; The wind curtailment power during the t-schedule period; This is the penalty coefficient for discarded light; The power of abandoned optical signals during scheduling period t; This refers to the cost coefficient for energy storage charging and discharging losses. Let t be the absolute value of the energy storage charging and discharging power during the scheduling period t; The unit carbon emission cost coefficient; The total carbon emissions during scheduling period t; For insufficient rotational reserve penalty coefficient; The required spinning reserve capacity of the system during the t-schedule period; The actual spinning reserve capacity of the system during the t-schedule period; Adjusting the wear coefficient for thermal power units; For the change in power output of thermal power units during the dispatch period t; This is the energy storage lifespan loss coefficient; The depth of energy storage discharge during time period t.

4. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 3, characterized in that, The optimized scheduling model also includes a knowledge base, which is used to store the operation of updating the optimized scheduling model based on the post-evaluation results using a multi-strategy fusion method. When the optimization scheduling model is used to perform optimization calculations based on multi-source data of the current-time wind-solar-thermal-storage integrated system, the parameters of the optimization scheduling model are dynamically adjusted through the knowledge base, including: Based on the optimized scheduling model update operation, scene clustering is performed using the maximum likelihood estimation method and the expectation-maximization algorithm to obtain the basic scene data; Based on the basic data of the scenario, the parameter change trend is obtained by prediction through an LSTM network. The overall benefit improvement rate is calculated based on the parameter change trend and the optimized scheduling model update operation. Based on the aforementioned comprehensive benefit improvement rate, a dynamic adjustment strategy for the optimized scheduling model is generated using the knowledge distillation method. The parameters of the optimized scheduling model are dynamically adjusted using a dynamic adjustment strategy.

5. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 1, characterized in that, The actual operating results include actual wind and solar power output and actual load data; Based on the actual operation results, predictive operation data is generated using a multi-scenario virtual operation model, including: Based on the actual operating data, virtual independent operating data is calculated using the independent operating data of wind, solar, thermal and energy storage from historical data. Based on the actual operating data, virtual joint operation data is obtained by optimizing the scheduling model. Based on the actual operating data, virtual ideal operating data is obtained by offline optimization calculation using a global optimal algorithm.

6. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 5, characterized in that, The post-evaluation results include quantitative benefit indicators and correlation evaluation results; The dimensions for quantifying benefits include: economic benefits, new energy consumption, model accuracy, energy storage health, environmental benefits, and comprehensive benefits. The economic benefit dimension includes the absolute value index of synergistic efficiency, the relative value index of synergistic efficiency, and the optimization potential index. The new energy consumption includes consumption increase indicators and consumption rate indicators; The model accuracy includes simulation deviation rate and decision consistency index; The energy storage health includes cycle life loss indicators and utilization efficiency indicators; The environmental benefits include carbon emission reduction indicators and carbon emission intensity indicators; The benefit quantification indicators are calculated from the actual operating data and the predicted operating data using the parameter relationships of the benefit quantification dimensions, including: The absolute value index of synergistic efficiency is calculated based on virtual independent operation data and actual operation data. The calculation formula is: ； in, This represents the total cost of virtual independent operations in the virtual independent operation data. This refers to the actual total operating cost in the actual operating data; Based on the absolute value index of synergistic efficiency and the virtual independent operation data, calculate the relative value index of synergistic efficiency. The calculation formula is: ； Calculate the optimization potential index based on the virtual ideal operating data and the actual operating data. The calculation formula is: ； in, The total cost of running a virtual ideal system based on virtual ideal system operating data; Calculate the absorption capacity improvement index based on the virtual independent operation data and actual operation data. The calculation formula is: ； in, This refers to the total amount of energy wasted by virtual independent operations in the virtual independent operation data; This refers to the total amount of energy wasted during actual operation, as shown in the actual operational data. The absorption rate index is calculated based on the actual operating data. The calculation formula is: ； in, This refers to the total amount of available new energy sources in the actual operational data; The simulation deviation rate index is calculated based on virtual joint operation data and actual operation data. The calculation formula is: ； in, The total cost calculated for the virtual joint operation model; Calculate the decision consistency index based on the virtual joint operation data and the actual operation data. The calculation formula is: ； in, For virtual joint operation contribution plans in virtual joint operation data; To contribute to the actual execution of the actual operating data; Calculate cycle life loss index based on the actual operating data. The calculation formula is: ； in, This is the loss coefficient; For exponential coefficients; The discharge depth of the i-th cycle in the actual operating data; The utilization efficiency index is calculated based on the actual operating data and the virtual independent operating data. The calculation formula is as follows: ； in, This refers to the actual total discharge in the actual operating data; The rated energy storage capacity in the virtual independent operating data; Carbon emission reduction targets are calculated based on virtual independent operation data and actual operation data. The calculation formula is: ； in, This refers to the carbon emissions of virtual independent operations (VIOs) within the VIO data. This refers to the actual carbon emissions from actual operation in the operational data. Carbon emission intensity index calculated based on actual operational data The calculation formula is: ； in, This refers to the actual total discharge in the actual operating data; The comprehensive benefit index is calculated based on the aforementioned synergistic efficiency relative value index, absorption rate index, simulation deviation rate index, cycle life loss index, and carbon emission intensity index to determine the comprehensive benefit dimension. The calculation formula is: ； in, This is the economic benefit weighting coefficient; The weighting coefficient for absorption benefits; These are the model accuracy weighting coefficients; This is the weighting coefficient for energy storage lifespan; Environmental benefit weighting coefficient; This is the preset baseline carbon emission intensity.

7. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 6, characterized in that, Based on the aforementioned benefit quantification indicators, correlation evaluation results are generated through trend analysis and anomaly detection methods, including: Based on the aforementioned benefit quantification indicators, the data characteristics of the benefit quantification indicators are calculated using a sliding window mechanism; The operational trend of the decision-making system is based on the data characteristics of the aforementioned benefit quantification indicators; Based on the aforementioned data characteristics, anomalies in the quantification of benefits are identified using a cumulative sum algorithm. Based on the system's operational trends and abnormal changes, correlation evaluation results are generated by calculating the correlation coefficient between the benefit quantification indicators and external factors.

8. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 1, characterized in that, The multi-strategy fusion method includes a rule base parameter fine-tuning strategy, a fuzzy logic multi-parameter collaborative adjustment strategy, a reinforcement learning self-optimization strategy, and a Bayesian optimization hyperparameter tuning strategy.

9. The closed-loop optimization scheduling method for a combined wind, solar, thermal, and energy storage system according to claim 8, characterized in that, Based on the post-evaluation results, the scheduling model is updated and optimized using a multi-strategy fusion method, including: Based on the aforementioned benefit quantification indicators and preset error thresholds, a feedback adjustment method for optimizing the scheduling model is determined through a hierarchical triggering mechanism. The feedback adjustment method includes a model adjustment method and a parameter adjustment method. The model adjustment method updates and optimizes the scheduling model algorithm using a multi-strategy fusion method based on the post-evaluation results; the parameter adjustment method updates and optimizes the scheduling model parameters using a multi-strategy fusion method based on the post-evaluation results.

10. A closed-loop optimization scheduling system for a wind-solar-thermal-storage integrated system based on post-evaluation feedback, characterized in that, Including storage media and processor; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the closed-loop optimization scheduling method for the wind-solar-thermal-storage integrated system as described in any one of claims 1 to 9.

Images