Intelligent charging decision-making method and system for photovoltaic storage and charging integrated equipment

Abstract

The invention discloses an intelligent charging decision-making method and system for photovoltaic storage and charging integrated equipment, and relates to the technical field of charging decision-making, and the method comprises the steps: constructing a decision-making agent based on a double-end decision-making, which comprises a power grid end and a load end, by aiming at target optical storage equipment and combining a network access scene and constructing a state space through source-network-load-storage-control; the charging decision task is interacted, and the charging decision task is a long and short time task and is marked with a task intention; and the charging decision task is imported into a decision agent, non-dominated sorting decision optimization and double-end strategy registration correction under double-end decision are executed in parallel by initializing a state space, a charging strategy is determined, and charging management and control are performed in response to the target optical storage equipment. The technical problem that in the charging process of the photovoltaic power storage integrated equipment in the prior art, the integrating degree of intelligent decision and actual application scenes is insufficient is solved, and the technical effect of improving the intelligent level of the charging strategy is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging decision-making, and particularly to an intelligent charging decision-making method and system for a photovoltaic storage and charging integrated device. Background Art

[0002] With the rapid development of new energy and the wide application of photovoltaic power generation technology, photovoltaic energy storage integrated devices are more and more widely used in energy storage and management. However, with the continuous expansion of the scale of photovoltaic power generation and the complexity of the power grid load, the monitoring and management of the charging process of photovoltaic energy storage integrated devices face huge challenges. Traditional photovoltaic energy storage charging control methods often have problems such as insufficient optimization of charging strategies, and it is difficult to meet the requirements of modern power systems for high efficiency, accuracy, and adaptability. Summary of the Invention

[0003] This application provides an intelligent charging decision-making method and system for a photovoltaic storage and charging integrated device, which is used to solve the technical problem of insufficient fit between intelligent decision-making and actual application scenarios in the charging process of a photovoltaic energy storage integrated device in the prior art.

[0004] In view of the above problems, this application provides an intelligent charging decision-making method and system for a photovoltaic storage and charging integrated device.

[0005] In the first aspect of this application, an intelligent charging decision-making method for a photovoltaic storage and charging integrated device is provided. The method includes: For a target photovoltaic energy storage device, in combination with the grid connection scenario, a state space is built with source-grid-load-energy storage-control, a decision-making agent based on dual-end decision-making is constructed, which includes a grid side and a load side; an interactive charging decision-making task, where the charging decision-making task is a long-term and short-term task and is marked with a task intention; the charging decision-making task is imported into the decision-making agent, and by initializing the state space, non-dominated sorting decision optimization under dual-end decision-making and dual-end strategy registration correction are executed in parallel to determine a charging strategy and perform charging control in response to the target photovoltaic energy storage device.

[0006] In the second aspect of this application, an intelligent charging decision-making system for a photovoltaic storage and charging integrated device is provided. The system includes: The decision-making agent construction module is used to build a state space with source-grid-load-storage-control for the target photovoltaic energy storage device in combination with the grid connection scenario, and construct a decision-making agent based on dual-terminal decision-making, where the dual terminals include the grid side and the load side; the charging decision task interaction module is used to interact with the charging decision task, where the charging decision task is a long-term and short-term task marked with task intention; the charging control module is used to import the charging decision task into the decision-making agent, and by initializing the state space, parallelly execute non-dominated sorting decision optimization and dual-terminal strategy registration and correction under dual-terminal decision-making to determine the charging strategy and perform charging control in response to the target photovoltaic energy storage device.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application is directed to a target photovoltaic energy storage device, combines the grid connection scenario, builds a state space with source-grid-load-storage-control, and constructs a decision-making agent based on dual-terminal decision-making, which includes the grid side and the load side; interacts with the charging decision task, where the charging decision task is a long-term and short-term task marked with task intention; imports the charging decision task into the decision-making agent, and by initializing the state space, parallelly executes non-dominated sorting decision optimization and dual-terminal strategy registration and correction under dual-terminal decision-making to determine the charging strategy and perform charging control in response to the target photovoltaic energy storage device. The present invention solves the technical problem that the existing technology has insufficient compatibility between intelligent decision-making and actual application scenarios during the charging process of photovoltaic energy storage integrated devices. By constructing a decision-making agent based on dual-terminal decision-making, combining the source-grid-load-storage-control state space, and adopting non-dominated sorting decision optimization and dual-terminal strategy registration and correction, the optimization and dynamic adjustment of the charging strategy are realized, achieving the technical effect of improving the intelligent level of the charging strategy. Description of the Drawings

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings without creative efforts based on these drawings.

[0009] Figure 1 It is a schematic flow chart of the intelligent charging decision-making method for the photovoltaic energy storage and charging integrated device provided in the embodiment of this application.

[0010] Figure 2 It is a schematic structural diagram of the intelligent charging decision-making system for the photovoltaic energy storage and charging integrated device provided in the embodiment of this application.

[0011] Explanation of the reference numerals: Decision-making agent construction module 11, charging decision task interaction module 12, charging control module 13. Specific Embodiments

[0012] The present application provides an intelligent charging decision-making method and system for a photovoltaic storage and charging integrated device. Aiming at solving the technical problem that the existing technology has insufficient fit between intelligent decision-making and actual application scenarios during the charging process of a photovoltaic storage integrated device, by constructing a decision-making intelligent agent based on dual-terminal decision-making, combining the source-network-load-storage-control state space, and adopting non-dominated sorting decision optimization and dual-terminal strategy registration correction, the optimization and dynamic adjustment of the charging strategy are realized, achieving the technical effect of improving the intelligent level of the charging strategy.

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts belong to the scope protected by the present application.

[0014] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0015] Embodiment 1, as Figure 1 shown, the present application provides an intelligent charging decision-making method for a photovoltaic storage and charging integrated device, and the method includes: Step S100: For a target photovoltaic storage device, in combination with the grid connection scenario, build a state space with source-network-load-storage-control, and construct a decision-making intelligent agent based on dual-terminal decision-making, where the dual-terminal includes the grid side and the load side.

[0016] In the embodiment of the present application, first, for a target photovoltaic storage device, in combination with a specific grid connection scenario, a source-network-load-storage-control state space is constructed. This state space comprehensively models various key factors of the photovoltaic storage device, including the output of the photovoltaic power source (source), the state and demand of the grid side (network), the power demand of the load side (load), the charge / discharge state of the energy storage device (storage), and the control mechanism (control). By building this state space, the mutual relationships between various factors in the system can be clearly captured, providing rich data support for subsequent decision-making.

[0017] Next, with the power grid side as the dominant, a first decision-making unit is constructed. In this unit, the charging strategy of the energy storage and photovoltaic (PV) devices is mainly determined by the operating requirements of the power grid, focusing on optimizing the interaction between the energy storage and PV devices and the power grid to ensure the stability of operating parameters such as the frequency and voltage of the power grid. The decision-making of the power grid side will dynamically adjust the charging and discharging process of the energy storage device according to the real-time changes in the power grid load, ensuring the stable operation of the power grid and optimizing the balance of the power grid load.

[0018] Then, with the load side as the dominant, a second decision-making unit is constructed. The goal of this unit is to coordinate the relationship between the energy storage and PV devices and the load demand. When the load demand fluctuates, by adjusting the charging and discharging strategy of the energy storage device, while ensuring that the load demand is met, the power transmission loss is reduced as much as possible, and the energy utilization efficiency of the system is improved.

[0019] These two decision-making units are parallel in structure and are coordinated through decision fusion rules. The role of the decision fusion rules is to integrate the decisions of the power grid side and the load side, eliminate strategy conflicts by coordinating the goals of both ends, ensure their collaborative optimization, and thus achieve the optimal energy scheduling and stable operation of the overall system.

[0020] Finally, through supervised training, these two decision-making units continuously learn and adjust to form the final decision-making intelligent agent. Through continuous training, the decision-making intelligent agent can adapt to different grid connection scenarios and adjust the charging strategy in real time according to the changes in the power grid load and the state of the energy storage device.

[0021] Furthermore, in the method provided by the application embodiment, constructing a decision-making intelligent agent based on dual-end decision-making further includes: Constructing a state space with source-grid-load-energy storage-control, with the power grid side as the decision-making dominant to construct a first decision-making unit; constructing a state space with source-grid-load-energy storage-control, with the load side as the decision-making dominant to construct a second decision-making unit; paralleling the first decision-making unit and the second decision-making unit at the same level, introducing decision fusion rules, and supervising and training the decision-making intelligent agent.

[0022] In the application embodiment, first, a source-grid-load-energy storage-control state space is constructed to integrate the key factors in the operation of the energy storage and PV devices. This step collects data of the photovoltaic power source (source), power grid (grid), load (load), energy storage device (energy storage), and control mechanism (control) to build a comprehensive multi-dimensional model. Specifically, the output power of the photovoltaic power source, the fluctuation of the load demand, the frequency and voltage of the power grid, and the charging and discharging state of the energy storage device are all included in the state space. This information provides real-time data support for subsequent decision optimization, enabling the decision-making intelligent agent to understand the key states of the system in real time.

[0023] Next, a first decision-making unit is constructed, with the power grid side as the dominant. In this unit, an optimization algorithm is adopted to adjust the charging and discharging strategies of energy storage devices in real time according to the power grid requirements. By inputting the real-time data of the power grid, such as the load, frequency, and voltage of the power grid, and using an optimization algorithm (such as linear programming), the optimal charging and discharging strategies of the energy storage devices are calculated. The core of this decision-making unit is to ensure the balance and stability of the power grid during the charging process by optimizing the scheduling of the power grid load, avoid grid overload or unstable states, and improve the energy coordination efficiency between the power grid and energy storage devices.

[0024] Subsequently, based on the same source-network-load-energy storage-control state space, a second decision-making unit is constructed, with the load side as the decision-making dominant. In this unit, a reinforcement learning method is adopted to optimize the charging and discharging strategies of energy storage devices in an adaptive manner. Reinforcement learning learns from the load changes and guides the agent to select the optimal charging and discharging strategies through a reward mechanism. When the load demand is high, the agent discharges to reduce the burden on the power grid; when the load demand is low, the agent charges to store electrical energy. Through training with historical data (such as historical load data and energy storage device operation data), the decision-making is continuously optimized, enabling the agent to make reasonable responses in the face of changing load demands.

[0025] These two decision-making units are at the same level and make independent decisions from the power grid side and the load side respectively. To achieve coordination between the power grid side and the load side, a decision fusion rule is introduced. This rule fuses the decision results of the power grid side and the load side, adopts a multi-objective optimization method (such as Pareto optimization), eliminates possible conflicts, and ensures that the decisions of the two ports can work in coordination. By adjusting the strategies of the two sides through an optimization algorithm, the charging and discharging processes of the photovoltaic energy storage devices can achieve the optimal energy scheduling and balance under the dual constraints of the power grid and load demands.

[0026] Finally, the entire decision-making agent is optimized through supervised training. During the training process, the decision-making agent is trained with historical data (including photovoltaic power generation output, power grid load changes, energy storage device charging and discharging data, etc.). By calculating the gap between the predicted results and the actual execution results of the model, the parameters of the model are adjusted to gradually improve the decision-making strategy. This process uses the backpropagation algorithm to update the parameters. Through multiple iterations of training, the charging decision-making ability of the decision-making agent is optimized. After supervised training, the decision-making agent can flexibly respond to the changes in the power grid and load in the actual operating environment and provide efficient and stable charging decisions.

[0027] Furthermore, the method provided by the application embodiment further includes: For the state space, extract the loss characteristics and uncertainty characteristics based on the elements in the source-grid-load-energy storage-control. Among them, the loss characteristics at least include line loss and machine loss; according to the loss characteristics and the uncertainty characteristics, perform two-terminal decoupling based on the grid side and the load side, and perform decision compensation on the mapped first decision unit and the second decision unit.

[0028] In the embodiment of the present application, first, based on the analysis of the source-grid-load-energy storage-control state space, extract the loss characteristics and uncertainty characteristics of each element in this space. The loss characteristics include line loss and machine loss. Line loss refers to the energy loss caused by the resistance when the current flows through the transmission line during the power transmission process. Usually, by measuring the magnitudes of the current and voltage and using Ohm's law to calculate the loss in the power transmission line. In addition, machine loss refers to the energy loss caused by internal mechanical friction, resistance, etc. when the energy storage device and other related devices are operating. Such losses can be estimated through device efficiency measurement and thermodynamic models. By extracting these loss characteristics, it can provide the necessary data support for the subsequent optimization of the charging strategy, help reduce energy waste, and improve the charging efficiency.

[0029] Next, analyze the uncertainty characteristics, which stem from external changing factors such as weather changes and load fluctuations. For example, the output power of photovoltaic power generation is closely related to weather conditions, and the fluctuation of light intensity will affect the photovoltaic power generation. Similarly, the load demand will also fluctuate due to electricity consumption patterns, seasonal changes, or emergencies (such as equipment failures). To effectively cope with these uncertainties, the Monte Carlo simulation method is adopted. This method generates different weather scenarios and load demand curves through random sampling, simulates the system behavior under different environments, so as to predict possible future changes and provide a basis for decision-making.

[0030] After clarifying the loss characteristics and uncertainty characteristics, perform two-terminal decoupling. Two-terminal decoupling means separately optimizing the decision-making processes on the grid side and the load side and solving them independently. The goal of the grid side is to maintain the stability of the grid. Therefore, factors such as grid load, frequency, and voltage need to be considered, and an optimal charging and discharging strategy is solved using an optimal scheduling algorithm (such as linear programming) to avoid grid overload or imbalance. The load side focuses on solving how to optimize the charging and discharging process of the energy storage device to ensure that the load demand is met. Through the load prediction algorithm, the change of the load can be accurately predicted, and the charging and discharging scheduling of the energy storage device can be carried out based on the load demand.

[0031] After the decisions at the grid side and the load side are independently optimized, the decision compensation stage is entered. The purpose of decision compensation is to ensure that the decisions at both ends can be coordinated. Specifically, when conflicts may occur in the decisions at the two ports (for example, the grid side requires more charging while the load side hopes to reduce the load), balance can be achieved through weight adjustment. Weight adjustment is achieved by assigning different priorities to the grid side and the load side. In actual operation, the stability of the grid side is usually more important than the load demand, so a higher decision weight is assigned to the grid side. If there are conflicts in the decisions at both ends, the needs of the grid side are given priority according to the adjusted weights, and the decisions of the load side are adjusted through appropriate strategies so that the goals of both are finally balanced.

[0032] Through this series of steps, the charging strategies at the grid side and the load side are independently optimized, and through the decision compensation mechanism, coordination between the two is ensured. Finally, by reducing losses and reasonably coping with uncertainties, the charging process can achieve efficient and stable operation and can adapt to changes in grid and load demands.

[0033] Step S200: Interact with the charging decision task, where the charging decision task is a long-term and short-term task and is marked with a task intention.

[0034] In the embodiment of the present application, the charging decision task is obtained by interacting with a preset database. The charging decision task is a long-term and short-term task. The long-term task is used to cope with the grid load changes in a relatively long period, usually spanning several hours or longer; while the short-term task responds quickly to the load fluctuations within a short time, usually lasting from a few minutes to several hours. Each charging task has a clear task intention, that is, a charging goal, which may be charging (increasing the power of the energy storage device) or discharging (releasing power to the grid).

[0035] Step S300: Import the charging decision task into the decision agent, and by initializing the state space, parallelly execute the non-dominated sorting decision optimization under the dual-end decision and the dual-end strategy registration correction to determine the charging strategy and perform charging control in response to the target photovoltaic energy storage device.

[0036] In the embodiment of the present application, first, the charging decision task is imported into the decision agent, and then the state space is initialized. This process provides comprehensive information support for the decision agent by integrating the real-time data of the photovoltaic energy storage device (such as grid load, photovoltaic power generation, charging state of the energy storage device, etc.). Through initialization, preliminary data such as grid load and energy storage device state are obtained, providing a necessary starting point for subsequent decision optimization.

[0037] Then, non-dominated sorting decision optimization under two-terminal decision-making is executed in parallel. In this stage, the decisions at the grid side and the load side are optimized independently. The grid side mainly focuses on grid stability, while the load side optimizes the charging and discharging processes of energy storage devices. Non-dominated sorting decision optimization simultaneously optimizes multiple objectives through multi-objective optimization methods (such as NSGA-II), balancing the grid stability and the efficient operation of energy storage devices to ensure their collaborative optimization. On this basis, two-terminal strategy registration and correction are continued. This process coordinates and adjusts the decision-making strategies at the grid side and the load side to ensure that the two decision-making ports can work in harmony and avoid strategy conflicts. Through the correction of the strategies, it is ensured that the objectives at the grid side and the load side can be balanced in the charging decision-making, achieving a better energy dispatch.

[0038] Finally, after optimization and adjustment, the charging strategy is determined and its response is applied to the target photovoltaic energy storage device for charging control.

[0039] Furthermore, in the method provided by the application embodiment, determining the charging strategy further includes: Identifying the charging decision task, and determining the initial quantity and the target variable; initializing the state space according to the initial quantity to determine the initialized space; performing variable combination optimization based on the target variable for the initialized space to determine the charging strategy.

[0040] In the embodiment of the present application, first, the charging decision task is identified. By relying on information such as grid load, photovoltaic power generation, and the state of energy storage devices, the specific type of charging task is determined. The charging tasks are classified according to the task intention (such as charging or discharging), and the goal of each task is determined. This step provides the initial conditions for the subsequent optimization process by obtaining real-time grid load data, photovoltaic power generation prediction data, and energy storage device state data to clarify the specific requirements of the task.

[0041] Next, the initial quantity and the target variable are determined. This step mainly obtains the known initial quantities from the grid load, photovoltaic power generation prediction, and the current state of energy storage devices, such as photovoltaic power generation and grid load demand. These known quantities serve as fixed values in the state space, providing a starting point for subsequent decision optimization. According to the task goal (such as charging or discharging), the optimization target variable is determined based on the changes in these initial quantities, such as grid stability and the charging and discharging efficiency of energy storage devices.

[0042] Then, the state space is initialized. This step forms a complete operation framework by inputting the known initial quantities (such as photovoltaic power generation and grid load) into the model of the state space. Each element (source, grid, load, storage, control) in the state space is set according to the task requirements and known data to ensure that the starting state of the system can provide an effective data basis for optimization. Through initialization, a basic environment for optimization is obtained.

[0043] Next, perform variable combination optimization based on the target variables. This step uses a multi-objective optimization method to optimize the grid-side target and the load-side target respectively. First, identify the task intention, clarify whether the charging task is charging or discharging, and decouple the task. For the grid-side target, optimize the charging strategy required for grid stability to ensure grid load balance. For the load-side target, optimize the charge and discharge strategy of the energy storage device to ensure meeting the load demand and optimizing the energy storage efficiency. Each target is adjusted through a variable optimization method, and finally an optimal strategy is determined for each port (grid side, load side).

[0044] Finally, determine the charging strategy according to the two optimal strategies. This step combines the optimization strategies of the grid side and the load side, and through the coordination of the two, finally obtains the best charging strategy.

[0045] Furthermore, in the method provided by the application embodiment, when performing variable combination optimization based on the target variables, it further includes: Identify the task intention, decouple the task intention with the grid side and the load side to determine the first task target and the second task target; taking the first task target as the guide, perform variable combination optimization based on the target variables on the initialization space to determine the first strategy; taking the second task target as the guide, perform variable combination optimization based on the target variables on the initialization space to determine the second strategy; determine the charging strategy according to the first strategy and the second strategy.

[0046] In the embodiment of the present application, first identify the task intention. This process clarifies the goal of the charging task according to the grid load demand, the photovoltaic power generation amount, and the charge and discharge demand of the energy storage device. For example, the task target of the grid side may be to reduce the peak-valley difference of the load, that is, to optimize the charge and discharge process of the energy storage device to keep the grid load balanced and avoid excessive peak loads or too low valley loads. The task target of the load side may be to reduce the charging cost. By optimizing the charging time and the charge and discharge strategy, charging is carried out as much as possible during the period with lower electricity prices, thereby reducing the electricity cost.

[0047] Then decouple the task intention with the grid side and the load side, that is, divide the goal of the task into the grid-side target and the load-side target respectively. Among them, the grid-side target is the first task target, and the load-side target is the second task target. The grid-side target usually focuses on reducing the peak-valley difference of the load, optimizing the charge and discharge strategy of the energy storage device, balancing the grid load, and ensuring the stability of the grid. The load-side target focuses on reducing the charging cost, that is, charging during the electricity price trough, and reducing the electricity bill expenditure by optimizing the charging time. By decoupling the task target, optimize for the grid and the load respectively to ensure that the two goals are independently and effectively achieved.

[0048] Then, with the grid - side target as the guide, perform variable - combination optimization based on target variables on the initialized space. This process is optimized for the grid - side target through a multi - objective optimization method. During this optimization process, consider multiple factors such as grid load, energy storage device status, and photovoltaic power generation, and find the optimal charging strategy by adjusting the combination of these variables. This process ensures that the energy storage device can efficiently discharge to balance the grid load, and finally obtains the first strategy, that is, the optimal charging strategy for the grid side.

[0049] Subsequently, with the load - side target as the guide, perform variable - combination optimization based on target variables on the initialized space. This optimization process is similar to the grid - side optimization, but the focus shifts to the load - side target, that is, reducing the charging cost. By adjusting the charging timing and strategy of the energy storage device, ensure charging during the low - electricity - price period to minimize the charging cost. Finally, this process will obtain the second strategy, that is, the optimal charging strategy for the load side.

[0050] Finally, determine the charging strategy according to the first strategy and the second strategy. At this time, map the grid - side strategy (such as the charge - discharge operation for stabilizing the grid) and the load - side strategy (such as reducing the charging cost), and find the first registered part and the second unregistered part. For the second unregistered part, introduce a cost function and perform balanced adjustment based on the grid and load - side targets to ensure the optimality of the charging strategy. Through this process, finally determine a charging strategy that comprehensively considers grid demand and load demand, ensuring that the energy storage device can be flexibly adjusted under different demands.

[0051] Furthermore, in the method provided by the application embodiment, determining the charging strategy according to the first strategy and the second strategy further includes: Map the first strategy and the second strategy to determine the first registered part and the second unregistered part; for the second unregistered part, introduce a cost function to perform balanced adjustment based on the first task target and the second task target, and determine the charging strategy.

[0052] In the embodiments of the present application, first, the first policy and the second policy are mapped. This process compares the charging policies at the grid side and the load side to identify the common and different parts between them. Specifically, the first policy is a charging policy obtained based on grid-side objectives (such as reducing the peak-to-valley difference of the load), while the second policy is a policy obtained based on load-side objectives (such as reducing charging costs). To perform this step, a decision fusion method is used to compare and analyze the optimization policies at the grid side and the load side, and identify the consistent part of the optimal decisions of both. That is, the first registered part is the part of the policy that both consider the best, which represents the part where both the grid stability and economic objectives are met. The second non-registered part is the part where there are differences in objectives between the two, which may be the case where the grid side needs more charging to balance the grid load, while the load side tends to reduce charging costs.

[0053] Next, for the second non-registered part, a cost function is introduced for adjustment. The cost function is a mathematical expression that measures the difference between the grid-side and load-side objectives. This function takes into account grid stability (such as grid load, frequency stability, etc.) and the economic objectives of the load side (such as the impact of electricity price fluctuations on the charging policy). For example, when the grid side requires more energy storage discharge to meet grid demand, the load side may need to adjust its charging timing to charge when the electricity price is low, thereby reducing electricity costs. The cost function calculates the conflict or difference between the two objectives and gives a numerical value representing the optimization degree of the current policy. The policy is adjusted according to the output of this cost function to ensure the balance between the grid-side and load-side objectives.

[0054] Finally, the charging policy is determined. This process combines the optimization results of the first registered part and the second non-registered part to generate a comprehensive charging policy. Specifically, according to the adjusted policy, the optimal charging decision is finally determined. This process uses a multi-objective optimization algorithm, such as Pareto optimization, to ensure that in the case of multi-objective conflicts, the policies at the grid side and the load side can achieve the best balance. At this stage, based on information such as grid load, photovoltaic power generation, and charging costs, the final charging policy is generated to ensure that the energy storage device can find the optimal charging and discharging strategy between grid demand and load economic requirements, thereby improving charging efficiency and reducing grid fluctuations.

[0055] Furthermore, the method provided by the application embodiments further includes: Determining policy registration based on mapping consistency; wherein, the part where the first policy and the second policy are mapped consistently is used as the first registered part.

[0056] In the embodiments of the present application, policy registration determination is performed based on mapping consistency. First, the charging policies at the grid end and the load end are compared to find the parts that match in the charging decision. Specifically, the first policy is the charging policy obtained based on grid-end goals (such as reducing the peak-valley difference of the load), and the second policy is the charging policy obtained based on load-end goals (such as reducing the charging cost). Through policy comparison and analysis, the decision-making policies at the grid end and the load end are compared one by one to find the parts that are considered optimal by both, and these parts are the parts with consistent mapping. That is, both the grid end and the load end believe that charging or discharging operations during a specific time period can achieve the optimal charging decision, and this policy simultaneously meets the grid demand and the load demand. These common parts are called the first registration part, that is, the best policy part that is consistent between the grid end and the load end.

[0057] Further, in the method provided by the embodiments of the application, when performing variable combination optimization on the initialization space based on the target variable, it further includes: Determine an initial policy set, where each initial policy is marked with a fitness value; stratify the initial policy set according to the non-dominance relationship, calculate the crowding degree of the policies in the non-dominated layer and select the parent policies to continue crossover and mutation, and select the first iterative policy set according to the non-dominated sorting and the crowding degree; use the first iterative policy set to perform iteration until the convergence condition is met, and determine the first policy according to the fitness value.

[0058] In the embodiments of the present application, first, an initial policy set is determined. This process is achieved by generating a set of charging policies, considering factors such as grid load, photovoltaic power generation, and energy storage device status. Each charging policy is preliminarily set according to this information, and the fitness value of each policy is calculated. The calculation of the fitness value uses a simple evaluation method. First, it is calculated by comparing the performance of the charging policy under three main goals (grid load balance, charging efficiency, and charging cost optimization). For example, the peak-valley difference of the grid load, charging efficiency, and charging cost can be calculated and normalized, and finally, a weighted average is performed according to the preset weights of the goals to obtain the comprehensive fitness value of each policy.

[0059] Next, the initial policy set is stratified according to the non-dominance relationship, and this process uses non-dominated sorting. The non-dominated sorting method evaluates whether each policy is dominated by other policies. If a policy is not worse than other policies in all goals, then it is considered non-dominated and enters the same layer. Non-dominated sorting hierarchically divides the policies into multiple layers. The first layer contains the optimal policies, the second layer contains the policies that are not dominated in the first layer, and so on. In this way, the policies are identified and stratified for subsequent optimization.

[0060] Next, calculate the crowding degree of the strategies in the non-dominated layer. This step uses the crowding degree metric. The crowding degree metric evaluates the "density" of strategies by calculating the distance between each strategy and its neighboring strategies within the same layer. Strategies with lower crowding degrees are located at the edges of the layer, while strategies with higher crowding degrees are located in the middle of the layer. By calculating the crowding degree, it is possible to assess which strategies are in relatively isolated positions in the multi-objective space and which strategies may overlap with other strategies. This method ensures the diversity of the strategy set and avoids over-concentration or local optimal solutions during the optimization process.

[0061] Then, select the parent strategies to continue with crossover and mutation. This process uses the binary tournament selection method. This method selects the strategies with higher fitness as parents to enter the crossover and mutation phases by comparing the fitness of the strategies. In the tournament selection method, multiple pairs of strategies are randomly selected for comparison, and the optimal strategy is selected to enter the next round of crossover and mutation operations. The crossover operation exchanges some features of the two parent strategies, while the mutation operation randomly changes certain strategy parameters to increase the diversity of the strategy set. Through this step, while maintaining the diversity of the strategies, new possible solutions are explored to further improve the fitness of the strategies.

[0062] Next, select the first iteration strategy set according to non-dominated sorting and crowding degree. After crossover and mutation, use the non-dominated sorting and crowding degree calculation methods to screen the newly generated strategies. Through non-dominated sorting, select those strategies with higher fitness and excellent performance under multiple objectives, and through crowding degree calculation, ensure the diversity in the strategy set. Finally, after screening, form the first iteration strategy set, and this set of strategies will enter the next round of iteration.

[0063] Finally, iterate with the first iteration strategy set until the convergence condition is met. During this process, through continuous iteration, operations such as crossover and mutation, non-dominated sorting, and crowding degree calculation are performed to gradually optimize the strategy set. After each round of iteration, check whether the convergence condition is met (such as the fitness change tends to be stable or the maximum number of iterations is reached). If the condition is met, stop the iteration. Finally, select the optimal charging strategy, that is, the first strategy, from the iteration process through the fitness value.

[0064] Furthermore, in the method provided by the application embodiment, after responding to the target optical storage device, it further includes: Determine the expected response based on the state space according to the charging strategy; with the expected response as a reference, trace the execution chain of the charging strategy to determine the execution response; take the difference between the expected response and the execution response to locate the element features based on the state space, and perform the charging feedback control of the target optical storage device.

[0065] In the embodiments of the present application, according to the charging strategy, the expected response based on the state space is first determined. This process predicts the execution effect of the charging strategy by analyzing factors such as grid load, photovoltaic power generation, and the current charging state of the energy storage device. The changes in grid load and the state of the energy storage device will affect the charge and discharge operations of the energy storage device. Therefore, by examining known data such as grid load and photovoltaic power generation, the charge or discharge operations that the energy storage device should perform at different time periods are inferred. For example, the expected response may indicate that the energy storage device should start discharging during peak grid load and should be charged during off-peak grid load.

[0066] Next, with the expected response as a reference, the execution chain of the charging strategy is actually monitored, that is, the reactions generated by the energy storage device during the actual charging process are traced. This process obtains data such as the charging current, discharge current, and grid load of the energy storage device through real-time monitoring to form an execution response. The execution response refers to the behavior of the energy storage device during actual operation, and these behaviors reflect the actual results during the execution of the charging strategy. Through real-time monitoring, the execution of each step is tracked to ensure that the charge and discharge operations of the energy storage device conform to the expected charging strategy.

[0067] Next, the difference between the expected response and the execution response is calculated. This step evaluates the execution effect of the charging strategy by calculating the difference between the two. To quantify the difference, a simple difference calculation is used to obtain the deviation between the two, such as by obtaining the absolute error or difference between the expected value and the actual value. This difference reflects the degree of deviation between the actual execution effect of the charging strategy and the expected effect. Through this difference calculation, the possible problems or parts that do not meet the expectations in the charging strategy are identified, providing a direction for improvement.

[0068] Then, the element features based on the state space are located, that is, through the difference calculation, the key factors causing the deviation are analyzed. At this time, according to the deviation of the charging strategy, it is judged whether the deviation is caused by the deviation of grid load prediction, the fluctuation of the efficiency of the energy storage device, or other factors (such as the change in photovoltaic power generation). By analyzing the source of the error, the key elements affecting the execution effect of the charging strategy are found and adjusted accordingly. For example, if it is found that the large fluctuation of grid load affects the execution effect, consider adjusting the timing of charge and discharge to reduce the deviation.

[0069] Finally, perform charging feedback control. This step adjusts the charging strategy based on the previously calculated error. By making feedback adjustments according to the difference between the actual execution response and the expected response, the optimization of the energy storage device charging process is ensured. By adjusting the charging or discharging power of the energy storage device, the deviation occurring during the execution process is corrected, ensuring that the energy storage device can be flexibly adjusted according to the changes in the grid load and charging cost, maintaining the stability of the grid and reducing the charging cost. This process is usually achieved through a feedback adjustment mechanism, which adjusts the charging strategy in real time according to the feedback information to achieve the best charging effect.

[0070] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects: The present application targets the target optical storage device, combines the grid connection scenario, constructs a state space with source-grid-load-storage-control, constructs a decision-making intelligent body based on two-terminal decision-making, including the grid side and the load side; interacts with the charging decision-making task, where the charging decision-making task is a long-term and short-term task marked with task intentions; imports the charging decision-making task into the decision-making intelligent body, initializes the state space, and concurrently executes non-dominated sorting decision optimization and two-terminal strategy registration correction under two-terminal decision-making to determine the charging strategy and perform charging control in response to the target optical storage device. The present invention solves the technical problem that the existing technology has insufficient fit between intelligent decision-making and actual application scenarios during the charging process of photovoltaic energy storage integrated devices. By constructing a decision-making intelligent body based on two-terminal decision-making, combining the source-grid-load-storage-control state space, and adopting non-dominated sorting decision optimization and two-terminal strategy registration correction, the optimization and dynamic adjustment of the charging strategy are realized, achieving the technical effect of improving the intelligent level of the charging strategy.

[0071] Embodiment 2, based on the same inventive concept as the intelligent charging decision-making method for the photovoltaic energy storage and charging integrated device in the foregoing embodiment, as Figure 2 shown, the present application provides an intelligent charging decision-making system for a photovoltaic energy storage and charging integrated device. The system and method embodiments in the present application are based on the same inventive concept. Among them, the system includes: A decision-making intelligent body construction module 11, which is used to construct a decision-making intelligent body based on two-terminal decision-making for the target optical storage device, combining the grid connection scenario, with a state space constructed by source-grid-load-storage-control, where the two terminals include the grid side and the load side; a charging decision-making task interaction module 12, which is used to interact with the charging decision-making task, where the charging decision-making task is a long-term and short-term task marked with task intentions; a charging control module 13, which is used to import the charging decision-making task into the decision-making intelligent body, initialize the state space, and concurrently execute non-dominated sorting decision optimization and two-terminal strategy registration correction under two-terminal decision-making to determine the charging strategy and perform charging control in response to the target optical storage device.

[0072] Furthermore, the system is also used to implement the following functions: Build a state space with source-network-load-storage-control, with the grid side as the decision-making dominant, and construct a first decision-making unit; build a state space with source-network-load-storage-control, with the load side as the decision-making dominant, and construct a second decision-making unit; juxtapose the first decision-making unit and the second decision-making unit at the same level, introduce decision fusion rules, and supervise and train the decision-making agent.

[0073] Furthermore, the system is also used to implement the following functions: For the state space, mine the loss characteristics and uncertainty characteristics based on the various elements within source-network-load-storage-control, where the loss characteristics at least include line loss and machine loss; according to the loss characteristics and the uncertainty characteristics, perform two-terminal decoupling based on the grid side and the load side, and perform decision compensation on the mapped first decision-making unit and second decision-making unit.

[0074] Furthermore, the system is also used to implement the following functions: Identify the charging decision task, determine the initial quantity and the target variable; according to the initial quantity, initialize the state space to determine the initialization space; for the initialization space, perform variable combination optimization based on the target variable to determine the charging strategy.

[0075] Furthermore, the system is also used to implement the following functions: Identify the task intention, decouple the task intention with the grid side and the load side to determine the first task target and the second task target; with the first task target as the guide, perform variable combination optimization based on the target variable on the initialization space to determine the first strategy; with the second task target as the guide, perform variable combination optimization based on the target variable on the initialization space to determine the second strategy; according to the first strategy and the second strategy, determine the charging strategy.

[0076] Furthermore, the system is also used to implement the following functions: Map the first strategy and the second strategy to determine the first registered part and the second unregistered part; for the second unregistered part, introduce a cost function to perform balance adjustment based on the first task target and the second task target to determine the charging strategy.

[0077] Furthermore, the system is also used to implement the following functions: Perform policy registration determination with mapping consistency; among them, the part where the first strategy and the second strategy are mapped consistently is used as the first registered part.

[0078] Furthermore, the system is also used to implement the following functions: Determine an initial policy set, where each initial policy is marked with a fitness value; stratify the initial policy set according to the non-dominated relationship, calculate the crowding degree of the policies in the non-dominated layer and select parent policies to continue crossover and mutation, and select the first iterative policy set according to non-dominated sorting and crowding degree; use the first iterative policy set to perform iterations until the convergence condition is met, and determine the first policy based on the fitness value.

[0079] Further, the system is also used to implement the following functions: According to the charging policy, determine the expected response based on the state space; with the expected response as a reference, trace the execution chain of the charging policy to determine the execution response; take the difference between the expected response and the execution response to locate the element features based on the state space, and execute the charging feedback control of the target optical storage device.

[0080] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of the present specification have been described. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0081] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.

[0082] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and variations.

Claims

1. An intelligent charging decision-making method for a photovoltaic energy storage and charging integrated device, characterized in that, The method includes: For the target optical storage device, in combination with the grid connection scenario, a state space is built with source-network-load-storage-control, and a decision-making intelligent agent based on dual-terminal decision-making is constructed, where the dual terminals include the grid side and the load side; Interact with the charging decision task, where the charging decision task is a long-term and short-term task and is marked with a task intention; Import the charging decision task into the decision-making intelligent agent. By initializing the state space, non-dominated sorting decision optimization under dual-terminal decision-making and dual-terminal strategy registration correction are executed in parallel to determine the charging strategy and perform charging control in response to the target optical storage device.

2. The intelligent charging decision-making method of the integrated photovoltaic energy storage and charging device according to claim 1, characterized in that Constructing a decision-making intelligent agent based on dual-terminal decision-making includes: Build a state space with source-network-load-storage-control, and with the grid side as the decision-making dominant, construct a first decision-making unit; Build a state space with source-network-load-storage-control, and with the load side as the decision-making dominant, construct a second decision-making unit; Arrange the first decision-making unit and the second decision-making unit side by side at the same level, introduce a decision-making fusion rule, and supervise and train the decision-making intelligent agent.

3. The intelligent charging decision-making method for the integrated photovoltaic energy storage and charging device according to claim 2, wherein, The method further includes: For the state space, mine the loss characteristics and uncertainty characteristics based on the various elements in source-network-load-storage-control, where the loss characteristics at least include line loss and machine loss; According to the loss characteristics and the uncertainty characteristics, perform dual-terminal decoupling based on the grid side and the load side, and perform decision-making compensation on the mapped first decision-making unit and the second decision-making unit.

4. The intelligent charging decision-making method of the integrated photovoltaic energy storage and charging device according to claim 1, characterized in that, Determining the charging strategy includes: Identify the charging decision task and determine the initial quantity and target variable; According to the initial quantity, initialize the state space to determine the initialized space; For the initialized space, perform variable combination optimization based on the target variable to determine the charging strategy.

5. The intelligent charging decision-making method of the integrated photovoltaic energy storage and charging device according to claim 4, characterized in that, Performing variable combination optimization based on the target variable includes: Identify the task intention, decouple the task intention with the grid side and the load side to determine the first task target and the second task target; Guided by the first task target, perform variable combination optimization based on the target variable on the initialized space to determine the first strategy; Guided by the second task target, perform variable combination optimization based on the target variable on the initialized space to determine the second strategy; According to the first strategy and the second strategy, determine the charging strategy.

6. The intelligent charging decision-making method for the integrated photovoltaic energy storage and charging device according to claim 5, wherein, According to the first strategy and the second strategy, determining the charging strategy includes: Map the first strategy and the second strategy to determine the first registered part and the second unregistered part; For the second unregistered part, introduce a cost function to perform an equilibrium adjustment based on the first task target and the second task target to determine the charging strategy.

7. The intelligent charging decision-making method for the photovoltaic energy storage and charging integrated device according to claim 6, wherein, Judge the strategy registration with mapping consistency; Among them, the part where the first strategy and the second strategy are mapped consistently is used as the first registered part.

8. The intelligent charging decision-making method for the integrated photovoltaic energy storage and charging device according to claim 6, characterized in that, Performing variable combination optimization based on the target variable on the initialized space includes: Determine the initial strategy set, where each initial strategy is marked with a fitness; Stratify the initial policy set according to the non-dominated relationship, calculate the crowding degree of the policies in the non-dominated layer, select the parent policies to continue crossover and mutation, and select the first iterative policy set according to the non-dominated sorting and crowding degree; Use the first iterative policy set to iterate until the convergence condition is met, and determine the first policy based on the fitness.

9. The intelligent charging decision-making method of the integrated photovoltaic energy storage and charging device according to claim 1, characterized in that, After responding to the target energy storage device, it includes: According to the charging policy, determine the expected response based on the state space; Taking the expected response as a reference, trace the execution chain of the charging policy to determine the execution response; Take the difference between the expected response and the execution response, locate the element features based on the state space, and execute the charging feedback control of the target energy storage device.

10. The intelligent charging decision-making system of the integrated photovoltaic energy storage and charging device is characterized in that, The system includes: A decision-making agent construction module, which is used for the target energy storage device, combines the grid connection scenario, builds a state space with source-network-load-storage-control, and constructs a decision-making agent based on double-end decision-making, where the double-end includes the grid side and the load side; A charging decision task interaction module, which is used to interact with the charging decision task, where the charging decision task is a long-term and short-term task and is marked with a task intention; A charging control module, which is used to import the charging decision task into the decision-making agent, and through initializing the state space, parallelly execute the non-dominated sorting decision optimization and double-end policy registration correction under double-end decision-making to determine the charging policy and perform charging control in response to the target energy storage device.

Images