A household energy storage collaborative optimization control method, device, equipment and storage medium

Abstract

The application discloses a kind of family energy storage collaborative optimization control method, device, equipment and storage medium.The method comprises: obtaining the operating state data and target income expectation of family energy storage network;Operating state data and target income expectation are input into family energy storage collaborative control model, and target collaborative control strategy is determined;Wherein, family energy storage collaborative control model includes: characteristic aggregation module, causal coding module, action mapping module and projection constraint module, characteristic aggregation module is used to carry out graph-level feature aggregation, determines global state vector and the node embedding feature of each energy node;Causal coding module is used to carry out causal strategy intention coding, determines global strategy intention vector;Action mapping module is used to determine the action scalar of each energy node;Projection constraint module is used to carry out power mapping, and determines target collaborative control strategy.The application can realize flexible and efficient control to family energy storage, guarantee the economy and robustness of energy storage control.

Description

Technical Field

[0001] This invention relates to the field of home energy storage control technology, and in particular to a method, device, equipment and storage medium for collaborative optimization control of home energy storage. Background Technology

[0002] In the fields of home energy management and smart grids, achieving precise and efficient coordinated control of distributed energy storage resources is a key technological means to improve grid operation stability, user-side economy and energy utilization efficiency, and has significant engineering application and market value.

[0003] Currently, traditional home energy storage control methods generally adopt a fixed rule-based operation mode. Operators need to pre-set the charging and discharging thresholds and timing strategies of the energy storage devices to generate and execute fixed control plans. However, traditional home energy storage control methods are highly dependent on preset rules, and the control strategies are rigid and lack adaptability. They are unable to handle real-time fluctuations in electricity price signals and dynamic changes in user electricity demand, resulting in poor economy, robustness, and interaction with the power grid. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and storage medium for coordinated optimization control of home energy storage, so as to achieve flexible and efficient control of home energy storage, reduce dependence on preset rules, greatly improve the flexibility and adaptability of control strategies, and can handle real-time fluctuating electricity price signals and dynamic changes in user electricity demand, thereby ensuring the economy, robustness, and interactivity of control strategies with the power grid.

[0005] According to one aspect of the present invention, a method for coordinated optimization control of home energy storage is provided, the method comprising: Obtain operational status data and expected returns for home energy storage networks; The operating status data and the expected target revenue of the home energy storage network are input into a pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, a target collaborative control strategy for the home energy storage network is determined, and the home energy storage network is collaboratively controlled based on the target collaborative control strategy. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

[0006] According to another aspect of the present invention, a home energy storage collaborative optimization control device is provided, the device comprising: The data acquisition module is used to acquire the operating status data and target revenue expectations of the home energy storage network. The collaborative control module is used to input the operating status data and the target revenue expectation of the home energy storage network into a pre-trained home energy storage collaborative control model for collaborative optimization, and determine the target collaborative control strategy of the home energy storage network based on the output of the home energy storage collaborative control model, so as to perform collaborative control of the home energy storage network based on the target collaborative control strategy. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the home energy storage collaborative optimization control method according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the home energy storage collaborative optimization control method according to any embodiment of the present invention.

[0009] The technical solution of this invention provides a comprehensive data foundation for subsequent energy storage control by acquiring the operational status data and target revenue expectations of a home energy storage network. The operational status data and target revenue expectations of the home energy storage network are input into a pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, a target collaborative control strategy for the home energy storage network is determined. This target collaborative control strategy is then used to collaboratively control the home energy storage network, thereby achieving intelligent, efficient, safe, and reliable collaborative control of home energy storage. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy. This invention acquires the operational status data and target revenue expectations of a home energy storage network, and combines them with a home energy storage collaborative control model to achieve flexible and efficient control of home energy storage. This reduces dependence on preset rules and greatly improves the flexibility and adaptability of the control strategy. It can handle real-time fluctuating electricity price signals and dynamic changes in user electricity demand, thereby ensuring the economy, robustness, and interactivity of the control strategy with the power grid.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a home energy storage collaborative optimization control method provided in Embodiment 1 of the present invention; Figure 2 This is a flowchart of a home energy storage collaborative optimization control method provided in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of a home energy storage collaborative optimization control device according to Embodiment 3 of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device that implements the home energy storage collaborative optimization control method of the present invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0014] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] Example 1 Figure 1This is a flowchart illustrating a method for coordinated optimization control of home energy storage according to Embodiment 1 of the present invention. This embodiment is applicable to situations involving coordinated optimization control of home energy storage. The method can be executed by a home energy storage coordinated optimization control device, which can be implemented in hardware and / or software. This home energy storage coordinated optimization control device can be configured in an electronic device. For example... Figure 1 As shown, the method includes: S110. Obtain the operating status data and target revenue expectations of the home energy storage network.

[0016] In this context, a home energy storage network refers to a collective network of energy devices within a home that can be uniformly dispatched and managed. This network may include energy storage devices (such as household batteries), adjustable loads (such as air conditioners, water heaters, and electric vehicle charging stations), and distributed power sources (such as photovoltaics), which are abstracted as interactive and controllable energy nodes. Operational status data refers to all relevant data describing the operational status of the home energy storage network at a given moment and its relationship to the external environment. Target revenue expectation refers to a control preference signal that can be set by the user or preset by the system, used to weigh multiple optimization objectives (such as economy and durability).

[0017] Specifically, dynamic equipment status and expected revenue are collected for home energy storage networks to obtain corresponding operational status data and target revenue expectations, providing a comprehensive data foundation for subsequent energy storage control.

[0018] For example, S110 may include: collecting device status data of the home energy storage network to obtain corresponding operating status data of the home energy storage network, wherein the operating status data includes: multi-dimensional state perception sequence, node network data and node constraint data, wherein the multi-dimensional state perception sequence is the electricity price signal, net household load power, photovoltaic output and weather forecast characteristics of the home energy storage network at different historical times; and determining the target revenue expectation of the home energy storage network based on a preset priority strategy, wherein the preset priority strategy includes: revenue priority strategy and lifetime priority strategy.

[0019] The multidimensional state-aware sequence refers to the time-series data set of a home energy storage network at different historical moments. It reflects the dynamic changes in the external environment and the system's own state, and can include: electricity price signals, net household load power, photovoltaic (PV) output, and weather forecast characteristics at different historical moments. Electricity price signals refer to past to present (and sometimes predicted future) electricity price information obtained from the power grid or electricity market. Net household load power refers to the real-time difference between the total household electricity load and the PV power generation (net load = total load PV output); a positive value indicates power intake from the grid, and a negative value indicates power feed into the grid. PV output refers to the real-time and historical power generation data of the home PV system. Weather forecast characteristics refer to weather forecast data such as irradiance, temperature, and cloud cover for a future period, used to predict trends in PV output and load changes. Node network data refers to data describing the physical topology and connectivity of the home energy storage network. Node constraint data refers to data describing the physical operating boundaries and capacity limitations of each energy device (node), which is the foundation for ensuring the safety and feasibility of control strategies. Preset priority strategies can refer to pre-set power consumption goals, such as maximizing economic benefits or maximizing equipment lifespan.

[0020] Specifically, a home energy management system can connect to the power grid or electricity market data interface to obtain electricity price signals for multiple historical periods. Through deployed smart meters or power sensors, the difference between the total household electricity load and photovoltaic (PV) power generation (net load = total load - PV output) can be monitored and calculated to obtain the household net load power for multiple historical periods. The PV inverter's monitoring interface can be used to obtain the PV system's output for multiple historical periods. Furthermore, by calling weather forecast APIs, meteorological forecast data regarding the impact of irradiance, temperature, humidity, cloud cover, and other factors on PV power generation and household load for a future period can be obtained. The system uses electricity price signals, household net load power, photovoltaic output, and weather forecast characteristics at different historical moments as a multi-dimensional state perception sequence for the home energy storage network. Each physical device in the home energy storage network (such as energy storage batteries, photovoltaic inverters, and adjustable loads controlled by smart sockets) is defined as a node. Based on the physical connections between devices, edges between nodes are defined to determine the node network data of the home energy storage network. The physical operating boundaries of each energy node are read or set from the technical specifications of each node device to obtain the node constraint data corresponding to the home energy storage network, thus providing a complete and multi-dimensional data foundation for subsequent intelligent decision-making. Users can select the desired priority strategy from preset priority strategies through the configuration interface to determine the expected target revenue of the home energy storage network. This allows user intentions to be clearly and intuitively conveyed to the control system, improving the system's interactivity and user experience.

[0021] S120. Input the operating status data and target revenue expectation of the home energy storage network into the pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, determine the target collaborative control strategy for the home energy storage network, and then perform collaborative control of the home energy storage network based on the target collaborative control strategy. The home energy storage collaborative control model includes: a feature aggregation module, a causal encoding module, an action mapping module, and a projection constraint module. The feature aggregation module is used to perform graph-level feature aggregation based on the operating status data to determine the global state vector of the home energy storage network and the node embedding features corresponding to each energy node. The causal encoding module is used to perform causal strategy intent encoding based on the operating status data and the global state vector to determine the global strategy intent vector of the home energy storage network. The action mapping module is used to perform modulation action mapping based on the target revenue expectation, the global strategy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module is used to perform power mapping based on the operating status data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

[0022] The home energy storage collaborative control model can refer to a core artificial intelligence model for intelligent collaborative optimization. It can be an end-to-end neural network comprising four modules: feature aggregation, causal encoding, action mapping, and projection constraints. Its function is to receive operational status data and target revenue expectations, calculate, and output collaborative control commands that satisfy safety constraints. The target collaborative control strategy can refer to the set of specific control commands that the model ultimately outputs and can be safely issued and executed. It is a set of vectors that specifies the specific physical power value that each energy node in the network should execute at the current moment (e.g., the battery is charged at 5kW, and the air conditioner power is set to 1.2kW). The global state vector can refer to a fixed-dimensional vector output by the feature aggregation module. It encodes the macroscopic comprehensive operational status of the entire home energy system at a certain moment by aggregating the features of all energy nodes. Node embedding features can refer to high-dimensional feature vectors representing individual energy devices and their relationships in the graph. The global strategy intent vector can refer to a high-dimensional vector representing the macroscopic control direction that the system should adopt under the current market and topology environment. The action scalar can refer to the normalized initial control quantity corresponding to a single energy node.

[0023] Specifically, the operational status data and target revenue expectations of the home energy storage network can be input into a pre-trained home energy storage collaborative control model. The feature aggregation module in the model performs global graph-level feature aggregation on the home energy storage network at the current moment based on the operational status data, determining the global state vector of the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent encoding based on the operational status data and the global state vector, determining the global policy intent vector of the home energy storage network. The action mapping module jointly decodes the global policy intent vector using the FiLM mechanism and the target revenue expectations, and performs modulated action mapping based on the obtained differentiated action modulation parameters of each node and the node embedding features corresponding to each energy node, determining the action scalar corresponding to each energy node. The projection constraint module performs physical power mapping and constraint projection based on the operational status data and the action scalar corresponding to each energy node, determining the target collaborative control strategy, and performing collaborative control of the home energy storage network according to the target collaborative control strategy. This achieves intelligent, efficient, safe, and reliable collaborative control of home energy storage, ensuring the economy and robustness of energy storage control.

[0024] In this embodiment, by acquiring the operational status data and target revenue expectations of the home energy storage network, a comprehensive data foundation is provided for subsequent energy storage control. The operational status data and target revenue expectations of the home energy storage network are input into a pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, a target collaborative control strategy for the home energy storage network is determined. This target collaborative control strategy is then used to collaboratively control the home energy storage network, thereby achieving intelligent, efficient, safe, and reliable collaborative control of home energy storage. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on operational state data to determine the global state vector of the home energy storage network and the node embedding features of each energy node. The causal coding module performs causal policy intent encoding based on operational state data and the global state vector to determine the global policy intent vector of the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features of each energy node to determine the action scalar of each energy node. The projection constraint module performs power mapping based on the operational state data and the action scalar of each energy node to determine the target collaborative control strategy. This invention, by acquiring the operational state data and target revenue expectation of the home energy storage network and combining them with the home energy storage collaborative control model, achieves flexible and efficient control of home energy storage, reduces dependence on preset rules, and greatly improves the flexibility and adaptability of the control strategy. It can handle real-time fluctuating electricity price signals and dynamic changes in user electricity demand, thereby ensuring the economy, robustness, and interaction with the power grid of energy storage control.

[0025] For example, the training process of the home energy storage collaborative control model in this invention can be as follows: An offline solver is used to generate a set of expert trajectories containing extreme price scenarios. A multi-objective reward function is defined: in, For time step The total reward function value; This represents the set of all energy nodes in the topology graph at the current moment. For time steps Real-time electricity price forecasts for the electricity market; For nodes At time step The execution power, a positive value represents discharging / selling electricity, and a negative value represents charging / buying electricity; To control the time step of the cycle (e.g., 15 minutes); The preset penalty weighting coefficient is used to balance the proportion of arbitrage profits with battery depreciation and loss of comfort. This is a cost function for battery life loss, used to quantify the impact of charge / discharge depth and power on lifespan; For time steps The state of charge of the battery; For time steps The battery charging and discharging power; Let this be the comfort loss function; These represent the current ambient temperature and the user-set ideal target temperature, respectively. It should be noted that during the offline training phase, the penalty weight coefficients in the multi-objective reward function... and The preferred value range is between 0.1 and 0.5.

[0026] Based on this reward function, a home energy storage collaborative control model is obtained through behavior cloning or offline reinforcement learning pre-training of the master policy network.

[0027] Example 2 Figure 2 This is a flowchart of a home energy storage collaborative optimization control method provided in Embodiment 2 of the present invention. Based on the above embodiments, this embodiment optimizes the step of "inputting the operating status data and target revenue expectation of the home energy storage network into a pre-trained home energy storage collaborative control model for collaborative optimization, and determining the target collaborative control strategy corresponding to the home energy storage network based on the output of the home energy storage collaborative control model". Explanations of terms that are the same as or corresponding to those in the above embodiments are not repeated here.

[0028] See Figure 2 Another home energy storage collaborative optimization control method provided in this embodiment specifically includes the following steps: S210. Obtain the operating status data and target revenue expectations of the home energy storage network.

[0029] S220. Input the operating status data into the feature aggregation module to perform graph-level feature aggregation, and determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node.

[0030] Specifically, the operational status data is input into the feature aggregation module. The feature aggregation module constructs the energy nodes in the home energy storage network into a graph structure and uses a graph neural network to extract node-level features. Combined with a global graph-level feature aggregation mechanism, it determines the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node, realizing the dynamic topology adaptation of the model. The model can adapt to scenarios with changes in the number of nodes (equipment commissioning and decommissioning) or changes in connection relationships, enhancing the robustness and flexibility of the system and providing reliable data input for subsequent steps.

[0031] For example, the feature aggregation module includes a graph structure generation unit and a feature aggregation unit. S220 may include: inputting the operating status data into the graph structure generation unit to construct the energy storage network and determine the energy node graph corresponding to the home energy storage network; inputting the energy node graph into the feature aggregation unit to extract node features, and combining the global graph-level feature aggregation mechanism to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node.

[0032] Among them, the energy node graph can refer to a dynamic graph data structure formed by abstracting and digitizing the home energy storage network.

[0033] Specifically, the operational status data is input to the graph structure generation unit, which reads the input node network data. Each energy storage device in the home energy storage network (such as a photovoltaic inverter, energy storage battery, adjustable load, etc.) is defined as a node in the graph. Based on the connection information in the node network data, the edges between nodes are defined to generate an energy node graph. The energy node graph is then input to the feature aggregation unit, which inputs the generated energy node graph into a pre-trained graph neural network for node-level feature extraction. This determines the node embedding features corresponding to each energy node. Combined with a global graph-level feature aggregation mechanism, the global state vector corresponding to the home energy storage network is determined, providing a solid information foundation for subsequent decisions of the entire control model.

[0034] For example, based on node network data, dynamically changing energy nodes are constructed as a graph structure. (Energy Node Graph), using graph neural networks to extract node-level features. (Node embedding features). A global graph-level feature aggregation mechanism is introduced: If global average pooling is used, the formula is: in, It is a global state vector with fixed dimensions, representing the current overall operating status of the entire house; This represents the total number of active nodes in the topology graph at the current moment. For nodes through The node embedding features extracted by layer graph convolution are high-dimensional feature vectors that integrate the node's own state and its neighborhood relationship information in the graph structure.

[0035] The global state vector As the topological condition input to the subsequent causal policy encoder, it is jointly generated with multidimensional state-aware sequence data to generate a fixed-dimensional global policy intent vector.

[0036] S230. Input the operating status data and global status vector into the causal coding module to perform causal strategy intent coding and determine the global strategy intent vector corresponding to the home energy storage network.

[0037] Specifically, the running state data and global state vector are input into the causal encoding module. The causal encoding module encodes the global state vector and the multi-state-aware sequences in the running state data into a fixed-dimensional global policy intent vector, thereby ensuring the spatial coordination and temporal foresight of policy generation in subsequent steps.

[0038] For example, the causal coding module includes a vector splicing unit and a causal coding unit. S230 may include: inputting the operating state data and the global state vector into the vector splicing unit for vector splicing to determine the target splicing vector; inputting the target splicing vector into the causal coding unit for causal strategy intent coding to determine the global strategy intent vector corresponding to the home energy storage network.

[0039] The target splicing vector can refer to the vector obtained by fusing the multi-dimensional time-series sensing sequence and the global state vector in the running state data.

[0040] Specifically, the operating status data and global status vector are input into the vector concatenation unit for vector concatenation to determine the target concatenation vector; the target concatenation vector is then input into the causal encoding unit for causal policy intent encoding using a pre-trained Transformer encoder with causal mask to generate the global policy intent vector corresponding to the home energy storage network, providing stable and reliable input data for the subsequent action mapping module.

[0041] For example, will With historical sequence Perform vector concatenation and input it into the Transformer policy encoder with causal masking: in, For a moment The global strategy intent vector (e.g., 64-dimensional) encodes the current macro market environment and topology operation status of the home energy network, serving as the global conditional input for subsequent FiLM conditional modulation. The global state vector has dimension . ; For a moment (Multidimensional state-aware sequence / vector); The length of the historical backtracking window (e.g., corresponding to 96 15-minute cycles in 24 hours). These are the trainable weight parameters of the causal Transformer policy encoder; This is a vector concatenation operation.

[0042] It should be noted that the causal mask ensures location Self-attention computation only aggregates positions Information that the model is strictly prohibited from being used at any time Access in reasoning Real-world future data ensures the causal validity of online reasoning. Output (like ) is a fixed-dimensional global strategy intent vector.

[0043] S240. Input the target profit expectation, the global policy intent vector and the node embedding features corresponding to each energy node into the action mapping module to perform modulated action mapping and determine the action scalar corresponding to each energy node.

[0044] Specifically, the target return expectation, the global policy intent vector, and the node embedding features corresponding to each energy node are input into the action mapping module. The action mapping module uses the FiLM mechanism to jointly decode the global policy intent vector with the target return expectation to obtain the differentiated action modulation parameters of each node. Based on the action modulation parameters and node embedding features corresponding to each energy node, node-level action decoding is performed to determine the action scalar corresponding to each energy node. This enables the unified macro policy to be personalized into differentiated and personalized control commands that adapt to the state of each node.

[0045] For example, the action mapping module includes a condition parameter generation unit and an action scalar determination unit. S240 may include: inputting the target revenue expectation and the global strategy intent vector into the condition parameter determination unit for node condition parameter modulation to determine the feature scaling parameter vector and feature offset parameter vector corresponding to each energy node in the home energy storage network; inputting the node embedding feature, feature scaling parameter vector and feature offset parameter vector corresponding to each energy node into the action scalar determination unit for decoding and normalization operations to determine the action scalar corresponding to each energy node.

[0046] The feature scaling parameter vector can be a vector used to determine the scaling factor of the values ​​in the corresponding dimension of the node embedding feature. The feature offset parameter vector can be a vector used to determine the offset of the values ​​in the corresponding dimension of the node embedding feature.

[0047] Specifically, the target revenue expectation and the global strategy intent vector are input to the condition parameter determination unit. The condition parameter determination unit jointly decodes the global strategy intent vector with the target revenue expectation through the FiLM mechanism to determine the feature scaling parameter vector and feature offset parameter vector corresponding to each energy node in the home energy storage network. The node embedding feature, feature scaling parameter vector and feature offset parameter vector corresponding to each energy node are input to the action scalar determination unit for node-level action decoding and normalization operations to determine the action scalar corresponding to each energy node, thereby achieving a precise mapping from collaborative decision-making to personalized execution.

[0048] For example, node-level condition parameters are generated using the FiLM mechanism: in, For nodes The feature scaling parameter vector and the feature offset parameter vector; This is the global policy intent vector; The preset target return expectation (strategy switching signal); : The trainable weight parameters of the conditional parameter generation network.

[0049] It should be noted that the scaling parameter of the conditional parameter generation network output is used when performing characteristic linear modulation. The numerical range is preferably mapped to [0.5, 2.0], with the offset parameter... The numerical range is preferably mapped to [-1.0, 1.0]. This range constraint ensures that the modulation intensity of the global intent on local features is within a controllable range, avoiding distortion of the control quantity due to drastic fluctuations in the expected target revenue (RTG value).

[0050] Node-level action decoding and normalization: , in, For nodes Modulated intermediate decision features; This is an element-wise multiplication operation (Hadamard product). For nodes Normalized action scalar (numerical) ).

[0051] S250: Input the operating status data and the action scalar corresponding to each energy node into the projection constraint module for power mapping and projection constraints, and determine the target cooperative control strategy.

[0052] Specifically, the operating status data and the action scalar corresponding to each energy node are input to the projection constraint module. The projection constraint module maps the action scalar of each node to the actual physical power upper and lower limits of the node through linear transformation to obtain the preliminary physical power command. The preliminary power command is input to a differentiable projection layer. The projection layer corrects the preliminary power command to the feasible region that satisfies all node constraint data (such as power limits and battery SoC dynamic equations) through mathematical operations, and outputs the final safe and executable target cooperative control strategy to ensure the safety and reliability of the control strategy.

[0053] For example, the projection constraint module includes a power mapping unit and a projection constraint unit. S250 may include: inputting the operating status data and the action scalar corresponding to each energy node into the power mapping module to perform physical power mapping on each energy node, and determining the candidate collaborative control strategy corresponding to the home energy storage network; inputting the candidate collaborative control strategy into the projection constraint unit for projection constraint correction, and determining the target collaborative control strategy.

[0054] Specifically, the operating status data and the action scalar corresponding to each energy node are input to the power mapping unit. The power mapping unit maps the action scalar of each node to the actual physical power upper and lower limits of that node through linear transformation, thereby obtaining the preliminary physical power command, i.e., the candidate cooperative control strategy. The candidate cooperative control strategy is then input to the projection constraint unit. The projection constraint unit corrects the preliminary power command to the feasible domain that satisfies the constraint data of all nodes (such as power limits and battery SoC dynamic equations) through mathematical operations, and outputs the final target cooperative control strategy that can be safely executed, ensuring the safety and reliability of the control strategy.

[0055] Example, power mapping: in, For action scalar Initial physical power control obtained by linear mapping For nodes The dynamic physical power upper and lower limits, which are sensed in real time at the current moment t, are determined by the device's own state (such as the battery BMS).

[0056] Initial power Projected onto the set of executable actions : in, The core SoC constraint formula involved is: In the feasible region In projection calculations, the lower bound of the SoC security envelope The value range is usually set as follows upper limit Set as .

[0057] in, To ensure absolute safety and feasibility, the physical control command vectors (target cooperative control strategy) that are finally issued to each device after being corrected by the differentiable projection layer are used. The mathematically feasible region is defined by the physical boundaries and operating rules (power limits, SoC dynamics, etc.) of all devices. The upper and lower limits of the safe battery capacity for operation; This represents the real-time state of charge of node (battery) n at the previous time (t-1); The conversion coefficient for the charge / discharge efficiency of node n; For energy nodes Rated electrical capacity.

[0058] It should be noted that after generating the target cooperative control strategy, the gradient of the projection layer can also be calculated using implicit function differentiation: in, To map vectors to the set of feasible regions The nonlinear projection function; the gradient value is used to propagate the error from the physical instruction layer back to the upstream policy network during the offline training phase, thereby achieving closed-loop optimization.

[0059] This gradient value serves as an error backpropagation mechanism during offline end-to-end training in the model training phase: it is used in a loss function defined by expert trajectory bias (behavioral cloning loss) or reinforcement learning reward signal. Starting with the gradient, the parameters are propagated back layer by layer through the Jacobian matrix of the projection layer to the power mapping layer, the FiLM conditional modulation layer, and the causal policy encoding layer, updating all network parameters. This mechanism ensures that the policy network continuously perceives the boundary of the physical feasible region during training, gradually learning to output initial actions that naturally satisfy the constraints without significant corrections, avoiding oscillations and divergence caused by hard truncation corrections during training. Without this gradient backpropagation path, the projection layer would become a dead end in the gradient path, the upstream network would not receive feedback on physical constraints, making it difficult to converge to a safe and optimal policy, and the end-to-end closed-loop optimization would fail.

[0060] The technical solution of this embodiment achieves dynamic topology adaptation of the model by inputting operational status data into a feature aggregation module for graph-level feature aggregation, determining the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The operational status data and global state vector are input into a causal coding module for causal policy intent encoding, determining the global policy intent vector corresponding to the home energy storage network, ensuring the spatial coordination and temporal foresight of subsequent policy generation steps. The target revenue expectation, global policy intent vector, and the node embedding features corresponding to each energy node are input into an action mapping module for modulated action mapping, determining the action scalar corresponding to each energy node, enabling the unified macro-strategy to be personalized into differentiated and individualized control commands adapted to the state of each node. The operational status data and the action scalar corresponding to each energy node are input into a projection constraint module for power mapping and projection constraints, determining the target collaborative control strategy and ensuring the safety and reliability of the control strategy. This invention, through the collaborative mechanism between the feature aggregation module, causal coding module, action mapping module, and projection constraint module, enables flexible and efficient control of home energy storage, greatly improving the flexibility and safety of the control strategy.

[0061] Example 3 Figure 3 This is a schematic diagram of a home energy storage collaborative optimization control device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310 and a collaborative control module 320; Among them, the data acquisition module 310 is used to acquire the operating status data and target revenue expectations of the home energy storage network; The collaborative control module 320 is used to input the operating status data and the target revenue expectation of the home energy storage network into a pre-trained home energy storage collaborative control model for collaborative optimization, and determine the target collaborative control strategy of the home energy storage network based on the output of the home energy storage collaborative control model, so as to perform collaborative control of the home energy storage network based on the target collaborative control strategy. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

[0062] In this embodiment, by acquiring the operational status data and target revenue expectations of the home energy storage network, a comprehensive data foundation is provided for subsequent energy storage control. The operational status data and target revenue expectations of the home energy storage network are input into a pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, a target collaborative control strategy for the home energy storage network is determined. This target collaborative control strategy is then used to collaboratively control the home energy storage network, thereby achieving intelligent, efficient, safe, and reliable collaborative control of home energy storage. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy. This invention acquires the operational status data and target revenue expectations of a home energy storage network, and combines them with a home energy storage collaborative control model to achieve flexible and efficient control of home energy storage. This reduces dependence on preset rules and greatly improves the flexibility and adaptability of the control strategy. It can handle real-time fluctuating electricity price signals and dynamic changes in user electricity demand, thereby ensuring the economy, robustness, and interactivity of the control strategy with the power grid.

[0063] Optionally, the data acquisition module 310 is specifically used to: collect the device status of the home energy storage network and acquire the corresponding operating status data of the home energy storage network, wherein the operating status data includes: multi-dimensional state perception sequence, node network data and node constraint data, wherein the multi-dimensional state perception sequence is the electricity price signal, net household load power, photovoltaic output and weather forecast characteristics of the home energy storage network at different historical times; Based on a preset priority strategy, the target expected revenue for the home energy storage network is determined. The preset priority strategy includes a revenue priority strategy and a lifespan priority strategy.

[0064] Optionally, the collaborative control module 320 includes: The feature aggregation layer is used to input the operating status data into the feature aggregation module for graph-level feature aggregation, and to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The intent encoding layer is used to input the operating status data and the global status vector into the causal encoding module to perform causal policy intent encoding, and determine the global policy intent vector corresponding to the home energy storage network. The action mapping layer is used to input the target revenue expectation, the global policy intent vector and the node embedding features corresponding to each energy node into the action mapping module to modulate the action mapping and determine the action scalar corresponding to each energy node. The collaborative control layer is used to input the operating status data and the action scalar corresponding to each energy node into the projection constraint module for power mapping and projection constraints, and to determine the target collaborative control strategy.

[0065] Optionally, the feature aggregation module includes a graph structure generation unit and a feature aggregation unit. The feature aggregation layer is specifically used for: inputting the operating status data into the graph structure generation unit to construct the energy storage network and determine the energy node graph corresponding to the home energy storage network; inputting the energy node graph into the feature aggregation unit to extract node features, and combining a global graph-level feature aggregation mechanism to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node.

[0066] Optionally, the causal coding module includes a vector concatenation unit and a causal coding unit. The intent coding layer is specifically used for: inputting the operating state data and the global state vector into the vector concatenation unit for vector concatenation to determine the target concatenation vector; and inputting the target concatenation vector into the causal coding unit for causal policy intent coding to determine the global policy intent vector corresponding to the home energy storage network.

[0067] Optionally, the action mapping module includes a condition parameter generation unit and an action scalar determination unit. Specifically, the action mapping layer is used to: input the target revenue expectation and the global strategy intent vector into the condition parameter determination unit for node condition parameter modulation, determining the feature scaling parameter vector and feature offset parameter vector corresponding to each energy node in the home energy storage network; and input the node embedding feature, feature scaling parameter vector, and feature offset parameter vector corresponding to each energy node into the action scalar determination unit for decoding and normalization operations, determining the action scalar corresponding to each energy node.

[0068] Optionally, the projection constraint module includes a power mapping unit and a projection constraint unit. The cooperative control layer is specifically used to: input the operating status data and the action scalar corresponding to each energy node into the power mapping module to perform physical power mapping on each energy node, and determine the candidate cooperative control strategy corresponding to the home energy storage network; input the candidate cooperative control strategy into the projection constraint unit for projection constraint correction, and determine the target cooperative control strategy.

[0069] The above-mentioned device can execute the home energy storage collaborative optimization control method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the home energy storage collaborative optimization control method.

[0070] Example 4 Figure 4 This is a schematic diagram of an electronic device implementing the home energy storage collaborative optimization control method of this invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0071] like Figure 4As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0072] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0073] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the home energy storage collaborative optimization control method.

[0074] In some embodiments, the home energy storage collaborative optimization control method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the home energy storage collaborative optimization control method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the home energy storage collaborative optimization control method by any other suitable means (e.g., by means of firmware).

[0075] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0076] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0077] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0078] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0079] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0080] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0081] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0082] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0083] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for coordinated optimization control of home energy storage, characterized in that, include: Obtain operational status data and expected returns for home energy storage networks; The operating status data and the expected target revenue of the home energy storage network are input into a pre-trained home energy storage collaborative control model for collaborative optimization. Based on the output of the home energy storage collaborative control model, a target collaborative control strategy for the home energy storage network is determined, and the home energy storage network is collaboratively controlled based on the target collaborative control strategy. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

2. The method according to claim 1, characterized in that, The acquisition of operational status data and target revenue expectations for the home energy storage network includes: The device status of the home energy storage network is collected to obtain the corresponding operating status data of the home energy storage network. The operating status data includes: multi-dimensional state perception sequence, node network data and node constraint data. The multi-dimensional state perception sequence is the electricity price signal, net household load power, photovoltaic output and weather forecast characteristics of the home energy storage network at different historical times. Based on a preset priority strategy, the target expected revenue for the home energy storage network is determined. The preset priority strategy includes a revenue priority strategy and a lifespan priority strategy.

3. The method according to claim 1, characterized in that, The step of inputting the operating status data and the expected target revenue of the home energy storage network into a pre-trained home energy storage collaborative control model for collaborative optimization, and determining the target collaborative control strategy for the home energy storage network based on the output of the home energy storage collaborative control model, includes: The operational status data is input into the feature aggregation module for graph-level feature aggregation to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The operating status data and the global status vector are input into the causal coding module to perform causal policy intent coding, thereby determining the global policy intent vector corresponding to the home energy storage network. The target revenue expectation, the global strategy intent vector, and the node embedding features corresponding to each energy node are input into the action mapping module to modulate the action mapping and determine the action scalar corresponding to each energy node. The operating status data and the action scalar corresponding to each energy node are input into the projection constraint module for power mapping and projection constraints to determine the target cooperative control strategy.

4. The method according to claim 3, characterized in that, The feature aggregation module includes a graph structure generation unit and a feature aggregation unit. The step of inputting the operating state data into the feature aggregation module for graph-level feature aggregation to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node includes: The operating status data is input into the graph structure generation unit to construct the energy storage network and determine the energy node graph corresponding to the home energy storage network. The energy node graph is input into the feature aggregation unit for node feature extraction, and combined with the global graph-level feature aggregation mechanism, the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node are determined.

5. The method according to claim 3, characterized in that, The causal coding module includes a vector concatenation unit and a causal coding unit. The step of inputting the operating state data and the global state vector into the causal coding module for causal policy intent coding to determine the global policy intent vector corresponding to the home energy storage network includes: The running status data and the global status vector are input into the vector splicing unit for vector splicing to determine the target splicing vector; The target concatenation vector is input into the causal coding unit for causal policy intent encoding to determine the global policy intent vector corresponding to the home energy storage network.

6. The method according to claim 3, characterized in that, The action mapping module includes a condition parameter generation unit and an action scalar determination unit. The step of inputting the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node into the action mapping module for modulating action mapping and determining the action scalar corresponding to each energy node includes: The target revenue expectation and the global strategy intent vector are input into the condition parameter determination unit to perform node condition parameter modulation, thereby determining the feature scaling parameter vector and feature offset parameter vector corresponding to each energy node in the home energy storage network. The node embedding features, feature scaling parameter vectors, and feature offset parameter vectors corresponding to each energy node are input to the action scalar determination unit for decoding and normalization operations to determine the action scalar corresponding to each energy node.

7. The method according to claim 3, characterized in that, The projection constraint module includes a power mapping unit and a projection constraint unit. The step of inputting the operating status data and the action scalar corresponding to each energy node into the projection constraint module for projection constraint correction and determining the target cooperative control strategy includes: The operating status data and the action scalar corresponding to each energy node are input into the power mapping module to perform physical power mapping on each energy node, thereby determining the candidate collaborative control strategy corresponding to the home energy storage network. The candidate cooperative control strategy is input into the projection constraint unit for projection constraint correction to determine the target cooperative control strategy.

8. A home energy storage collaborative optimization control device, characterized in that, include: The data acquisition module is used to acquire the operating status data and target revenue expectations of the home energy storage network. The collaborative control module is used to input the operating status data and the target revenue expectation of the home energy storage network into a pre-trained home energy storage collaborative control model for collaborative optimization, and determine the target collaborative control strategy of the home energy storage network based on the output of the home energy storage collaborative control model, so as to perform collaborative control of the home energy storage network based on the target collaborative control strategy. The home energy storage collaborative control model includes a feature aggregation module, a causal coding module, an action mapping module, and a projection constraint module. The feature aggregation module performs graph-level feature aggregation based on the operating state data to determine the global state vector corresponding to the home energy storage network and the node embedding features corresponding to each energy node. The causal coding module performs causal policy intent coding based on the operating state data and the global state vector to determine the global policy intent vector corresponding to the home energy storage network. The action mapping module performs modulation action mapping based on the target revenue expectation, the global policy intent vector, and the node embedding features corresponding to each energy node to determine the action scalar corresponding to each energy node. The projection constraint module performs power mapping based on the operating state data and the action scalar corresponding to each energy node to determine the target collaborative control strategy.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the home energy storage collaborative optimization control method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the home energy storage collaborative optimization control method according to any one of claims 1-7.

Images