Multi-objective optimization method for substation building light storage direct flexible system

Abstract

The present application relates to the technical field of electric energy storage system, disclose a substation building light storage direct flexible system multi-objective optimization method, comprising the following steps: first, the operation data of the substation building light storage direct flexible system and external environment data are obtained, and the subsequent optimization decision is provided with basic information;Subsequently, based on the block chain technology, the interaction behavior between at least two participants is evaluated by decentralization trust, and the collaborative coefficient related to global green power consumption in the system optimization target is dynamically corrected according to the trust evaluation result. By constructing a dynamic collaborative entropy field which unifies and quantifies the multidimensional value targets such as economy, environment and reliability, the complex substation building multi-objective optimization problem is ingeniously transformed into a minimum entropy field total entropy value problem with clear target, solving the problem that the prior art cannot effectively trade off and collaborative decision, and achieving the technical effect of overall and collaborative optimization of the system.

Description

Technical Field

[0001] This invention relates to the field of energy storage system technology, specifically a multi-objective optimization method for a building-based photovoltaic-storage-DC-flexible system in a substation. Background Technology

[0002] With the advancement of the "dual carbon" goals, the penetration rate of distributed renewable energy, represented by photovoltaics, in the building sector is increasing. For substations, which play a key role in the power grid, their auxiliary buildings also have superior conditions for installing distributed photovoltaics, configuring energy storage systems, applying DC power distribution and flexible loads. This has gradually formed a new type of energy system integrating power generation, energy storage, power consumption and flexible regulation, namely "photovoltaic-storage-DC-flexible". How to efficiently optimize and schedule such systems to achieve multiple goals of economy, environmental protection and reliability has become an important research topic.

[0003] However, existing technologies still have many shortcomings in optimizing the scheduling of photovoltaic-storage-DC-flexible systems in substations. Traditional optimization methods usually transform multi-objective problems into weighted single-objective problems for solution. However, the setting of weight coefficients is highly subjective and usually statically configured, making it difficult to dynamically adapt to complex operating conditions and changes in the external power grid environment. This makes it difficult for the system to achieve optimal comprehensive benefits in different scenarios. This static trade-off mechanism often fails to effectively handle the dynamic contradiction between local economic interests and the macro-level green energy consumption demand of the power grid. In addition, the robustness of existing methods also faces severe challenges. These methods generally adopt a two-stage model of "prediction first, optimization later". Their optimization results are highly dependent on the accuracy of prediction of uncertain factors such as photovoltaic output and load demand. In actual operation, prediction errors are unavoidable. Especially when facing sudden weather changes or load shocks, the scheduling strategy based on inaccurate predictions may become inapplicable or even harmful, thus seriously affecting the stability and economy of the system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a multi-objective optimization method for building-integrated photovoltaic-storage-DC-flexible systems in substations. This method solves the problems of insufficient overall system benefits and grid support capabilities in existing building-integrated photovoltaic-storage-DC-flexible systems during optimized scheduling, which are caused by difficulties in dynamically balancing multiple objectives, poor adaptability to external uncertainties, and lack of a reliable interaction mechanism for deep collaboration with the power grid.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multi-objective optimization method for a substation building-based photovoltaic-storage-DC-flexible system, comprising the following steps:

[0006] S1. First, obtain the operation data and external environment data of the photovoltaic-storage-DC-flexible system of the substation building to provide basic information for subsequent optimization decisions; then, based on blockchain technology, conduct decentralized trust assessment on the interaction behavior between at least two participants, and dynamically adjust the synergy coefficient related to the overall green electricity consumption in the system optimization target according to the trust assessment results.

[0007] S2. Construct a dynamic collaborative entropy field containing multi-dimensional value, transforming the multi-objective optimization problem of the substation building photovoltaic-storage-DC-flexible system into a problem of minimizing the total entropy value of the dynamic collaborative entropy field, wherein the modified collaborative coefficient is the key input parameter for constructing the dynamic collaborative entropy field;

[0008] S3. Using a multi-agent game learning framework, with the goal of minimizing the total entropy value of the dynamic collaborative entropy field, the substation building-based photovoltaic-storage-DC-flexible system is optimized and scheduled to generate the optimal scheduling strategy for the system; finally, the substation building-based photovoltaic-storage-DC-flexible system is controlled in real time according to the optimal scheduling strategy.

[0009] Preferably, the system acquires photovoltaic output data, energy storage system operation data, various load demand data, building operation environment data, and external environment data within the substation building. The external environment data includes grid price data, grid green electricity supply and demand status indication data, and meteorological forecast data. The acquired operation data and external environment data are then cleaned, missing value filled, outlier handled, and time synchronized.

[0010] Preferably, the blockchain-based approach specifically includes: constructing a distributed ledger network, whose participants include intelligent agents of the substation energy management system (…). ) and regional power grid dispatching and coordinating intelligent agents ( The participating parties encrypt their key operational data and interaction behavior data, and periodically report them to the distributed ledger via smart contracts; the smart contracts calculate the data based on the records in the ledger. Contribution reputation ( ) and the above Regarding the Trust score ;and based on the , and regional power grid green electricity supply and demand status indication Dynamically correct the synergy coefficient related to the contribution dimension of green energy consumption across the entire region in the system optimization objective. .

[0011] Preferably, the synergy coefficient ( The corrected formula is:

[0012]

[0013] in, Based on the synergy coefficient, These are the weighting coefficients.

[0014] Preferably, the photovoltaic power generation, available energy storage capacity, various electrical loads, heat loads, cold loads and adjustable potential of flexible loads in the substation building mentioned in step S2 are abstracted as generalized energy particles (GEPs).

[0015] Preferably, for each generalized energy particle At time step Calculate its entropy value The Value function in multiple dimensions and its corresponding dynamic synergy coefficient Weighted summation yields:

[0016]

[0017] in, It is a set of value dimensions.

[0018] Preferably, the multi-agent game-theoretic learning framework includes intelligent agents for the substation energy management system. Intelligent agents in uncertain environments ) and regional power grid dispatching and coordinating intelligent agents ), wherein The goal is to minimize the total entropy of the dynamic cooperative entropy field. The The goal is to maximize the Perceived The The goal is to minimize the curtailment of regional green energy and by directing it towards Release green electricity supply and demand signals to facilitate interaction.

[0019] Preferably, the multi-agent game learning framework learns in the following way:

[0020] The With the Perform adversarial reinforcement learning to update and optimize the [unclear text - possibly related to a specific technology or method]. The internal prediction model improves prediction accuracy and decision robustness against uncertainty; and, the With the To conduct collaborative reinforcement learning, the aforementioned Minimize itself Simultaneously, in conjunction with the trust assessment results obtained from the blockchain, the response is... The signal will be used to adjust the scheduling strategy.

[0021] Preferably, the reward function With the Inversely proportional:

[0022]

[0023] in, A positive reward coefficient.

[0024] Preferably, the optimal scheduling strategy obtained through multi-agent game learning is deployed to the energy management system (EMS) of the substation building; the EMS executes the optimal scheduling strategy based on real-time acquired data, the dynamically corrected coordination coefficient, and the current system state, and issues control commands to the execution unit of the photovoltaic-storage-DC-flexible system; and feeds back the actual system operation data to the data acquisition step, the trust assessment step, and the multi-agent game learning step to form a closed-loop continuous optimization process.

[0025] This invention provides a multi-objective optimization method for building-integrated photovoltaic-storage-DC-flexible systems in substations. It offers the following advantages:

[0026] 1. This invention constructs a dynamic collaborative entropy field that unifies and quantifies multiple value objectives such as economy, environment, and reliability. It cleverly transforms the complex multi-objective optimization problem of substation construction into a problem of minimizing the total entropy value of the entropy field with a clear objective. This solves the problem that existing technologies struggle to effectively weigh and make collaborative decisions when faced with multiple conflicting optimization objectives. It achieves the technical effect of holistic and collaborative optimization of the system, significantly improving the comprehensive benefits of energy utilization.

[0027] 2. This invention introduces a blockchain-based distributed trust assessment and adaptive correction mechanism for collaborative coefficients, establishing a transparent and tamper-proof interactive behavior recording and assessment system between the substation energy management system and the regional power grid dispatching system. This solves the problem of the difficulty in effectively coordinating local interests with the macro-goals of the power grid in traditional optimization methods. It achieves the technical effect of enabling the local optimization of substations to respond to the green electricity consumption demand of the power grid in real time and dynamically, and realizing deeper source-load interaction and win-win cooperation based on trust relationships.

[0028] 3. This invention employs a multi-agent game learning framework that includes an agent in an uncertain environment. It simulates external uncertainties such as photovoltaic output and load demand as an adversarial agent seeking to maximize the system entropy. This forces the energy management system agent to learn in an adversarial game, solving the problem of insufficient robustness of control strategies due to prediction errors in the traditional "prediction first, optimization later" method. It achieves the technical effect of significantly improving the decision robustness and operational reliability of the system in dealing with various uncertainties and emergencies in real and complex environments.

[0029] 4. This invention achieves integrated online optimization of prediction and decision-making by internalizing the prediction model into the reinforcement learning process of the intelligent agent of the energy management system. It solves the technical defects of traditional methods where the prediction module and the decision-making module are separated, which leads to the prediction error being transmitted and amplified step by step, affecting the final control effect. It achieves the technical effect of enabling the scheduling strategy and prediction capability to evolve synergistically, thereby improving the real-time performance, accuracy and overall operational efficiency of decision-making.

[0030] 5. This invention establishes a closed-loop feedback and continuous optimization process covering data collection, trust assessment, and game learning. It continuously feeds the actual operating data of the system back to each stage of the optimization method, solving the problem that traditional control strategies tend to become rigid once deployed and are difficult to adapt to system aging or long-term changes in the external environment. This achieves the technical effect of enabling the system to have long-term self-learning, self-adaptation, and self-optimization capabilities, thereby ensuring that it maintains its optimal operating state throughout its entire life cycle. Attached Figure Description

[0031] Figure 1 This is the main flowchart of the present invention;

[0032] Figure 2 This is a flowchart of the closed-loop optimization process of the present invention. Detailed Implementation

[0033] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see the appendix Figure 1 - Appendix Figure 2 This invention provides a multi-objective optimization method for a substation building-based photovoltaic-storage-DC-flexible system, comprising the following steps:

[0035] S1. First, obtain the operation data and external environment data of the building's photovoltaic-storage-DC-flexible system in the substation to provide basic information for subsequent optimization decisions; then, based on blockchain technology, conduct decentralized trust assessment on the interaction behavior between at least two participants, and dynamically adjust the synergy coefficient related to the overall green energy consumption in the system optimization objectives according to the trust assessment results.

[0036] Specifically, the following multi-source heterogeneous data is acquired in real time and periodically through a sensor network deployed inside the substation building and external data interfaces:

[0037] Photovoltaic (PV) data: including real-time photovoltaic output Historical power output data series, and solar irradiance predictions obtained through the meteorological service interface. and predicted ambient temperature ;

[0038] Storage data includes the real-time state of charge of the battery storage system. Health status and its maximum charge and discharge power and ;

[0039] Load data: including DC power load demand within substation buildings. AC power load demand Heat load demand Cooling load demand And identify and quantify the adjustability potential of the flexible load at the current moment. ;

[0040] Grid data: including real-time electricity purchase prices obtained from grid companies or the electricity market. Electricity sales price And the power grid green electricity supply and demand status indications issued by the regional power grid dispatch coordinator. and auxiliary service demand signals ;

[0041] Building operation data: including indoor temperature in key areas relative humidity and the operating status of major energy-consuming equipment. .

[0042] Data preprocessing: A series of processes are performed on the collected raw data to ensure data quality and consistency. This process includes cleaning the data to remove noise, filling missing values ​​using interpolation or prediction models, detecting and correcting outliers using statistical methods, and aligning and synchronizing data from different systems over time. All processed data will be normalized or standardized to meet the input requirements of subsequent models.

[0043] S2. Construct a dynamic collaborative entropy field containing multi-dimensional value. Transform the multi-objective optimization problem of the photovoltaic-storage-DC-flexible system in the substation building into the problem of minimizing the total entropy value of the dynamic collaborative entropy field. The modified collaborative coefficient is the key input parameter for constructing the dynamic collaborative entropy field. In step S2, the photovoltaic power generation, available energy storage capacity, various electrical loads, heat loads, cold loads and the adjustable potential of flexible loads in the substation building are abstracted as generalized energy particles (GEPs).

[0044] Specifically, a distributed ledger network based on consortium blockchain technology will be deployed, and core participants, such as intelligent agents of substation energy management systems, will be included. ) and regional power grid dispatching and coordinating intelligent agents ( As network nodes, each participant periodically shares its key operational data and interaction behavior data (e.g., The actual amount of green electricity consumed and the response actions to dispatch instructions; After the released green electricity supply and demand signals (such as those for electricity supply and demand) are encrypted, they are reported to the distributed ledger via smart contracts, forming a public, transparent, and tamper-proof record of behavior. Based on the real behavior data recorded on the chain, pre-deployed smart contracts automatically calculate the "trust score" and "contribution reputation" of each participant, for example, calculating... Contribution reputation ( )and right Trust score These indicators quantify the effectiveness and cooperation of all parties in promoting the regional consumption of green electricity. Based on real-time trust assessment results obtained from the blockchain and real-time green electricity supply and demand status indicators of the regional power grid .

[0045] S3. A multi-agent game learning framework is adopted to optimize the scheduling of the substation building-based photovoltaic-storage-DC-flexible system with the goal of minimizing the total entropy value of the dynamic collaborative entropy field, thereby generating the optimal scheduling strategy for the system. Finally, the substation building-based photovoltaic-storage-DC-flexible system is controlled in real time according to the optimal scheduling strategy.

[0046] Specifically, construct a game-theoretic learning framework involving at least three agents:

[0047] Substation energy management system intelligent agent ( Its goal is to minimize the total entropy of the dynamic cooperative entropy field by optimizing its own scheduling behavior. ;

[0048] Intelligent agents in uncertain environments ( Its goal is to maximize the potential of photovoltaic power generation by simulating external uncertainties such as photovoltaic output and load demand. Perceived .

[0049] Regional power grid dispatching and coordinating intelligent agent ( Its goal is to minimize the curtailment of regional green energy and to do so by directing energy towards green energy use. To cooperate in releasing green electricity supply and demand signals.

[0050] Learning through adversarial and cooperative game theory:

[0051] Adversarial learning: and Adversarial reinforcement learning is employed in game theory. Through continuous response The system generates various uncertainty scenarios, learns online and optimizes its internal prediction model, thereby improving the prediction accuracy and decision robustness for uncertainty.

[0052] Collaborative learning: and Conduct collaborative reinforcement learning. Minimize itself Simultaneously, by combining the trust assessment results obtained from the blockchain, a response is made. The signal will be used to adjust the scheduling strategy. reward function Its entropy Inversely proportional, that is ,in With a positive reward coefficient, through continuous game iteration and reinforcement learning algorithm training, it eventually converges and generates... Optimal scheduling strategy Finally, the optimal scheduling strategy obtained from the training is deployed into the actual energy management system to achieve real-time, intelligent, and coordinated control of the photovoltaic-storage-DC-flexible system resources, forming a closed-loop continuous optimization. This step specifically includes:

[0053] Strategy deployment: The training results obtained in step S4 Optimal scheduling strategy The model is deployed into the field energy management system (EMS) of the substation building as its core decision engine.

[0054] Real-time decision-making and command execution: at each time step The on-site EMS executes the optimal scheduling strategy based on the real-time data acquired (from step S1), the dynamically corrected coordination coefficient (from step S2), and the current system status. The system calculates and generates optimal control commands. These commands are then sent to execution units such as photovoltaic inverters, energy storage converters, and flexible load controllers to achieve real-time control of the system equipment.

[0055] Closed-loop feedback and continuous optimization: The actual operating data of the system is used as new input and fed back to step S1 to update the data basis, to step S2 to update the on-chain behavior records and use them for new trust assessment, and to step S4 as new learning experience to continuously iterate and optimize the agent's strategy, thus forming a closed-loop, adaptive and continuously optimized energy management process.

[0056] Acquire photovoltaic power output data, energy storage system operation data, various load demand data, building operation environment data, and external environment data within the substation building. External environment data includes grid price data, grid green electricity supply and demand status indication data, and weather forecast data. Clean the acquired operation data and external environment data, fill in missing values, handle outliers, and synchronize the time.

[0057] Specifically, acquiring operational data and external environmental data of the photovoltaic-storage-DC-flexible system in the substation building is achieved through sensor networks, metering instruments, communication interfaces, and external data service platforms deployed inside and outside the substation building.

[0058] Specifically, blockchain technology includes: constructing a distributed ledger network, whose participants include intelligent agents of the substation energy management system (…). ) and regional power grid dispatching and coordinating intelligent agents ( Participants encrypt their key operational and interactive data and periodically report it to the distributed ledger via smart contracts; the smart contracts then calculate... Contribution reputation ( )as well as right Trust score ;and based on , and regional power grid green electricity supply and demand status indication The synergy coefficient related to the contribution dimension of green energy consumption across the entire region in the dynamic correction system optimization objective is as follows: Coefficient of coordination ( The corrected formula for ) is:

[0059]

[0060] in, In time step Lower generalized energy particles Synergy coefficient in terms of contribution to green energy consumption across the entire region Based on the synergy coefficient, These are the weighting coefficients.

[0061] For each generalized energy particle At time step Calculate its entropy value , Value function in multiple dimensions and its corresponding dynamic synergy coefficient Weighted summation yields:

[0062]

[0063] in, A set of value dimensions Including the corrected synergistic coefficients from step S2; and aggregating the entropy values ​​of all generalized energy particles to obtain the total entropy value of the dynamic synergistic entropy field. Value Dimensions Set It should include at least economic efficiency, environmental friendliness, reliability, complementarity, and contribution to the absorption of green electricity across the entire region.

[0064] The multi-agent game theory learning framework includes intelligent agents in a substation energy management system. Intelligent agents in uncertain environments ) and regional power grid dispatching and coordinating intelligent agents ),in The goal is to minimize the total entropy of the dynamic cooperative entropy field. , The goal is to maximize Perceived , The goal is to minimize the curtailment of regional green energy and by directing it towards Interacting by releasing green electricity supply and demand signals, the multi-agent game learning framework learns through the following methods:

[0065] and Perform adversarial reinforcement learning to update and optimize in real time. Internal predictive models improve prediction accuracy and decision robustness against uncertainty; and, and Conduct collaborative reinforcement learning. Minimize itself Simultaneously, by combining the trust assessment results obtained from the blockchain, a response is made. The signal will be used to adjust the scheduling strategy. reward function and Inversely proportional:

[0066]

[0067] in, A positive reward coefficient.

[0068] The optimal scheduling strategy obtained through multi-agent game learning is deployed into the energy management system (EMS) of the substation building. Based on real-time acquired data, dynamically corrected coordination coefficients, and the current system state, the EMS executes the optimal scheduling strategy and issues control commands to the execution unit of the photovoltaic-storage-DC-flexible system. The actual system operation data is fed back to the data acquisition step, the trust assessment step, and the multi-agent game learning step to form a closed-loop continuous optimization process.

[0069] In summary, this invention provides a multi-objective optimization method for substation building photovoltaic-storage-DC-flexible systems. By constructing a dynamic collaborative entropy field that unifies and quantifies multiple value objectives such as economy, environment, and reliability, the complex multi-objective optimization problem of substation building is cleverly transformed into a problem of minimizing the total entropy value of the entropy field with a clear objective. This solves the problem of existing technologies struggling to effectively weigh and coordinate multiple conflicting optimization objectives, achieving the technical effect of holistic and collaborative system optimization and significantly improving the comprehensive benefits of energy utilization. Furthermore, by introducing a blockchain-based distributed trust assessment and adaptive correction mechanism for collaborative coefficients, a transparent and tamper-proof interactive behavior recording and evaluation system is established between the substation energy management system and the regional power grid dispatching system. This solves the problem of the difficulty in effectively coordinating local interests with the macro-level objectives of the power grid in traditional optimization methods, enabling the local optimization of the substation to respond in real-time and dynamically to the green electricity consumption needs of the power grid.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-objective optimization method for a substation building-based photovoltaic-storage-DC-flexible system, characterized in that, Includes the following steps: S1. First, obtain the operation data and external environment data of the photovoltaic-storage-DC-flexible system of the substation building to provide basic information for subsequent optimization decisions; then, based on blockchain technology, conduct decentralized trust assessment on the interaction behavior between at least two participants, and dynamically adjust the synergy coefficient related to the overall green electricity consumption in the system optimization target according to the trust assessment results. S2. Construct a dynamic collaborative entropy field containing multi-dimensional value, transforming the multi-objective optimization problem of the substation building photovoltaic-storage-DC-flexible system into a problem of minimizing the total entropy value of the dynamic collaborative entropy field, wherein the modified collaborative coefficient is the key input parameter for constructing the dynamic collaborative entropy field; S3. Using a multi-agent game learning framework, with the goal of minimizing the total entropy value of the dynamic collaborative entropy field, the substation building-based photovoltaic-storage-DC-flexible system is optimized and scheduled to generate the optimal scheduling strategy for the system; finally, the substation building-based photovoltaic-storage-DC-flexible system is controlled in real time according to the optimal scheduling strategy.

2. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, The system acquires photovoltaic output data, energy storage system operation data, various load demand data, building operation environment data, and external environment data within the substation building. The external environment data includes grid price data, grid green electricity supply and demand status indication data, and meteorological forecast data. The acquired operation data and external environment data are then cleaned, missing values ​​are filled, outliers are handled, and time is synchronized.

3. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, The blockchain-based technology specifically includes: constructing a distributed ledger network, whose participants include intelligent agents of the substation energy management system (…). ) and regional power grid dispatching and coordinating intelligent agents ( The participating parties encrypt their key operational data and interaction behavior data, and periodically report them to the distributed ledger via smart contracts; the smart contracts calculate the data based on the records in the ledger. Contribution reputation ( ) and the above Regarding the Trust score ;and based on the , and regional power grid green electricity supply and demand status indication Dynamically correct the synergy coefficient related to the contribution dimension of green energy consumption across the entire region in the system optimization objective. .

4. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 3, characterized in that, The synergy coefficient ( The corrected formula is: in, Based on the synergy coefficient, These are the weighting coefficients.

5. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, In step S2, the photovoltaic power generation, available energy storage capacity, various electrical loads, heat loads, cold loads, and adjustable potential of flexible loads within the substation building are abstracted as generalized energy particles (GEPs).

6. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, For each generalized energy particle At time step Calculate its entropy value The Value function in multiple dimensions and its corresponding dynamic synergy coefficient Weighted summation yields: in, It is a set of value dimensions.

7. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, The multi-agent game-theoretic learning framework includes intelligent agents in the substation energy management system. Intelligent agents in uncertain environments ) and regional power grid dispatching and coordinating intelligent agents ), wherein The goal is to minimize the total entropy of the dynamic cooperative entropy field. The The goal is to maximize the Perceived The The goal is to minimize the curtailment of regional green energy and by directing it towards Release green electricity supply and demand signals to facilitate interaction.

8. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 7, characterized in that, The multi-agent game-theoretic learning framework learns through the following methods: The With the Perform adversarial reinforcement learning to update and optimize the [unclear text - possibly related to a specific technology or method]. The internal prediction model improves prediction accuracy and decision robustness against uncertainty; and, the With the To conduct collaborative reinforcement learning, the aforementioned Minimize itself Simultaneously, in conjunction with the trust assessment results obtained from the blockchain, the response is... The signal will be used to adjust the scheduling strategy.

9. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 8, characterized in that, The reward function With the Inversely proportional: in, A positive reward coefficient.

10. The multi-objective optimization method for substation building-based photovoltaic-storage-DC-flexible system according to claim 1, characterized in that, The optimal scheduling strategy obtained through multi-agent game learning is deployed to the energy management system (EMS) of the substation building. The EMS executes the optimal scheduling strategy based on real-time acquired data, the dynamically corrected coordination coefficient, and the current system state, and issues control commands to the execution unit of the photovoltaic-storage-DC-flexible system. The actual system operation data is fed back to the data acquisition step, the trust assessment step, and the multi-agent game learning step to form a closed-loop continuous optimization process.

Images