Energy management method, system and device of distributed storage and charging system and storage medium

Abstract

The invention belongs to the technical field of power transformation and distribution, and discloses an energy management method, system and device for a distributed storage and charging system and a storage medium, and the method comprises the steps: constructing a machine networking architecture comprising a plurality of storage and charging units; when an external electric energy input or electric energy assistance demand exists in the machine networking architecture, calculating a corresponding competition value based on the energy storage residual electric quantity data and the predicted load corresponding to each storage and charging unit; and calculating an input distribution scheme of external electric energy and an electric energy assistance scheme between the storage and charging units based on the competition values corresponding to the storage and charging units. According to the technical scheme, the risk of system crash in a centralized control mode can be effectively avoided, meanwhile, the cost of the system is reduced, and the efficiency and stability of the system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power transformation and distribution, and in particular relates to an energy management method, system, device and storage medium for a distributed storage and charging system. Background Art

[0002] As the importance of energy storage systems in power systems becomes increasingly prominent, especially in the context of the rapid development of the electric vehicle industry, traditional large-scale centralized energy storage systems have gradually exposed some limitations. In order to meet these challenges, the distributed storage / charge system (DSCS) has become an emerging option for users due to its flexibility, rapid response and convenience. By combining the distributed storage / charge system with DC fast charging and super charging piles, and taking advantage of the shared DC bus characteristics of the storage / charge system, not only the working efficiency of the charging piles is improved, but also the demand for the capacity of dedicated transformer equipment for the charging piles is reduced.

[0003] However, most of the current distributed storage and charging system clusters adopt centralized control. Although this method can achieve overall coordination, it also has some disadvantages in practical applications, such as high requirements for central control equipment and transmission channels, increased risk of system crash as the number of equipment increases, and increased software and hardware costs.

[0004] At present, the decentralized distributed storage and charging system cluster coordination control scheme is adopted. Although the idea of ​​decentralization is proposed, it still uses a certain mechanism to determine a host to control the rest of the devices, and has not achieved the purpose of complete decentralization. The interval between the host failure and the election process will still cause the cluster to fail to work normally. At the same time, the cluster is only for the exchange of control commands, and does not coordinate the energy exchange of each storage and charging system unit through the DC bus. Therefore, the original problem of reduced energy utilization caused by transmission loss in the process of each unit directly sharing the DC bus has not been solved.

[0005] In the scenario of combining existing technologies with DC fast charging and super charging piles, although the operating efficiency of charging piles is improved by sharing the DC bus and the demand for the capacity of substation equipment is reduced, due to the lack of intelligent planning, the energy scheduling of each storage and charging unit when sharing the DC bus is often not optimized. This limitation will lead to uneven energy distribution, increased transmission losses, and difficulty in dynamically adjusting the charging load according to real-time needs, which will lead to the risk of system collapse and affect the energy efficiency of the overall system. Summary of the invention

[0006] The purpose of the present invention is to provide an energy management method, system, device and storage medium for a distributed storage and charging system to solve the problems existing in the above-mentioned prior art.

[0007] To achieve the above object, the present invention provides an energy management method for a distributed storage and charging system, comprising:

[0008] Step 1: Build a machine network architecture including several storage and charging units;

[0009] Step 2: When there is an external power input or power assistance demand in the machine networking architecture, a corresponding competition value is calculated based on the energy storage remaining power data and predicted load corresponding to each storage and charging unit;

[0010] Step 3: Calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units based on the competition value corresponding to each of the storage and charging units.

[0011] Optionally, the step 2 specifically includes:

[0012] Acquire training data, wherein the training data includes historical charging data, relative geographical location data and load data of each of the storage and charging units;

[0013] Construct a load forecasting model based on spatiotemporal graph convolutional network;

[0014] Based on the training data and in combination with the Grey Wolf Optimization Algorithm, the load forecasting model is trained and parameter optimization is performed to obtain an optimized load forecasting model;

[0015] Execute load prediction of each storage unit based on the optimized load prediction model to obtain predicted load data of each storage unit;

[0016] Obtaining the remaining energy storage capacity data and electricity price level data of each of the storage and charging units;

[0017] The energy storage remaining capacity data, electricity price level data and predicted load data of each storage and charging unit are weighted and summed to obtain a corresponding competitive value.

[0018] Optionally, training and parameter optimization of the load forecasting model based on the training data in combination with the Grey Wolf Optimization Algorithm specifically includes:

[0019] Extracting a hyperparameter combination included in the load forecasting model;

[0020] Based on the training data, the load forecasting model corresponding to each of the hyperparameter combinations is trained, and the training results are evaluated, and the evaluation results are used as the evaluation scores corresponding to each of the hyperparameter combinations;

[0021] Generating a gray wolf population for gray wolves using each of the hyperparameter combinations;

[0022] Based on the evaluation scores corresponding to the hyperparameter combinations and in combination with the gray wolf optimization algorithm, parameter optimization is performed within the gray wolf population to obtain the optimal hyperparameter combination;

[0023] An optimized load forecasting model is constructed based on the optimal hyperparameter combination.

[0024] Optionally, training the load forecasting model corresponding to each of the hyperparameter combinations based on the training data specifically includes:

[0025] Dividing the training data into a training set, a test set, and a validation set according to a preset ratio;

[0026] The load prediction model corresponding to each of the hyperparameter combinations is trained based on the training set, and the performance of the load prediction model corresponding to each of the hyperparameter combinations is verified and evaluated through the verification set, and the evaluation result is used as the evaluation score corresponding to each of the hyperparameter combinations.

[0027] Optionally, performing parameter optimization in the gray wolf population based on the evaluation scores corresponding to the hyperparameter combinations in combination with the gray wolf optimization algorithm specifically includes:

[0028] Setting the value range, population size and number of iterations of the hyperparameter combination, and using each of the hyperparameter combinations as gray wolves to generate the gray wolf population;

[0029] Based on the gray wolf optimization algorithm, each hyperparameter combination in the gray wolf population is iteratively updated multiple times. In each iterative update process, the evaluation score corresponding to each hyperparameter combination is calculated, and the gray wolf population is updated based on the evaluation score corresponding to each hyperparameter combination until the number of iterations reaches a preset value and the optimal hyperparameter combination is output.

[0030] Optionally, performing load forecasting of each storage unit based on the optimized load forecasting model specifically includes:

[0031] Extracting the time series characteristics and relative geographical location characteristics of the storage unit,

[0032] Performing modal decomposition on the time series features, combining the decomposed modal functions with relative geographical location features, and constructing a feature data set with spatiotemporal features;

[0033] The load forecasting model includes a first time convolution layer, a graph convolution layer, a second time convolution layer and a fully connected prediction layer connected in sequence, the feature data set is input into the load forecasting model, and features are extracted through each time convolution layer and graph convolution layer;

[0034] The extracted features are input into the fully connected prediction layer for classification prediction to obtain predicted load data.

[0035] Optionally, the external electric energy input distribution scheme and the electric energy assistance scheme between the storage and charging units are calculated based on the competition value corresponding to each of the storage and charging units, specifically including:

[0036] Input allocation scheme of external electric energy: sorting the competition values ​​corresponding to the storage and charging units, and selecting a number of storage and charging units as candidate storage and charging units in the sorting based on a preset competition threshold;

[0037] Based on the competition value corresponding to each candidate storage and charging unit, the input power of external electric energy is allocated in proportion to complete the external electric energy input allocation of the current cycle;

[0038] Electric energy assistance scheme between storage and charging units: the storage and charging unit to be assisted initiates an assistance request, and sorts the remaining storage and charging units except the storage and charging unit to be assisted based on the competition value, and selects the assisting storage and charging unit for assistance based on the sorting result.

[0039] An energy management system for a distributed storage and charging system, comprising:

[0040] A machine networking building module, used to build a machine networking architecture including several storage and charging units;

[0041] A competition mechanism module, used for calculating a corresponding competition value based on the energy storage remaining power data and predicted load corresponding to each of the storage and charging units when there is an external power input or power assistance demand in the machine networking architecture;

[0042] The assistance module is used to calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units according to the competition value corresponding to each of the storage and charging units.

[0043] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program so that the electronic device executes the energy management method of a distributed storage and charging system.

[0044] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the energy management method of a distributed storage and charging system.

[0045] The technical effects of the present invention are:

[0046] The present invention realizes energy sharing and information exchange between units through the hardware-level scheduling and response mechanism of the decentralized "machine networking" structure between storage and charging system units. The structure is compatible with the flexible networking mode of multiple units, reduces the deployment difficulty of the storage and charging system cluster, and expands its scope of application. The present invention effectively avoids the risk of system crash under centralized control mode through a decentralized control algorithm without a host, while reducing the cost of the system. Using the competition algorithm, the system can dynamically decide the subject of command issuance based on the total power of the power grid, the input requirements of each storage and charging system unit, and the load forecast value, thereby improving the reliability and flexibility of the system. The present invention further combines the deep learning algorithm to improve the scientific nature of the "machine networking" energy scheduling. By dynamically predicting the load of each unit, the system can provide a more accurate basis for competition weights when performing energy scheduling, thereby optimizing the overall energy management strategy and improving the efficiency and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0049] Figure 1 FIG. 1 is a diagram for assigning PS and PL functions of an EMS end of a storage unit in an embodiment of the present invention;

[0050] Figure 2 The structure diagram of the characteristic cables connecting the storage and charging units in the machine networking in the embodiment of the present invention;

[0051] Figure 3 A schematic diagram of a machine networking cluster structure in an embodiment of the present invention;

[0052] Figure 4 is a schematic diagram of a competition mechanism flow in an embodiment of the present invention;

[0053] Figure 5 is a specific calculation flow chart of the competition value in an embodiment of the present invention;

[0054] Figure 6 A flowchart of the specific steps of building a deep computing model using the VMD-GWO-ST-GCN algorithm in an embodiment of the present invention;

[0055] Figure 7It is a supercharging full service rate curve diagram of each strategy at different time periods in a daily scenario in an embodiment of the present invention;

[0056] Figure 8 This is a graph showing the ratio of the daily average amount of electricity used for service to the electricity cost under each strategy in the embodiment of the present invention. DETAILED DESCRIPTION

[0057] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0058] It should be understood that the terms described in the present invention are only for describing special embodiments and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that each intermediate value between the upper and lower limits of the scope is also specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the described range is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded in the scope.

[0059] Unless otherwise indicated, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art generally understood by the present invention. Although the present invention describes only preferred methods, any method similar or equivalent to that described herein may also be used in the implementation or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods related to the documents. In the event of a conflict with any incorporated document, the content of this specification shall prevail.

[0060] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention description without departing from the scope or spirit of the present invention. Other embodiments derived from the present invention description will be apparent to those skilled in the art. The present application description and examples are exemplary only.

[0061] The words “include,” “including,” “have,” “contain,” etc. used in this article are open-ended terms, meaning including but not limited to.

[0062] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0063] Embodiment 1

[0064] like Figure 1 - Figure 8 As shown, this embodiment provides an energy management method for a distributed storage and charging system, including:

[0065] Step 1: Build a machine network architecture including several storage and charging units;

[0066] Step 2: When there is an external power input or power assistance demand in the machine networking architecture, a corresponding competition value is calculated based on the energy storage remaining power data and predicted load corresponding to each storage and charging unit;

[0067] Step 3: Calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units based on the competition value corresponding to each of the storage and charging units.

[0068] The problem to be solved in this embodiment is: how to achieve efficient coordination and exchange of energy and information among multiple storage and charging system units, and to plan energy storage charging in an orderly manner according to the total power of the power grid and equipment requirements, and to avoid the risk of system collapse and cost increases caused by centralized control methods.

[0069] To solve the above problems, this embodiment proposes a decentralized "machine network" structure between storage and charging system units, which is compatible with the networking mode of 1-n units, and uses a competitive algorithm to dynamically decide the subject of issuing commands between nodes based on factors such as the total power input from the power grid, the input demand and geographical location of each storage and charging system unit, and the load prediction value calculated by the deep learning algorithm. A method for improving the scientific nature of "machine network" energy scheduling using a deep learning algorithm, dynamically predicting the load of each unit, and providing a basis for competitive weights during "machine network" energy scheduling.

[0070] This embodiment realizes energy sharing and information exchange between units through the hardware-level scheduling and response mechanism of the "machine networking" structure between decentralized storage and charging system units. This structure is compatible with the flexible networking mode of 1-n units, reduces the deployment difficulty of the storage and charging system cluster, and expands its scope of application. Through the decentralized control algorithm without a host, the risk of system crash under the centralized control mode is effectively avoided, while reducing the cost of the system. Using the competition algorithm, the system can dynamically decide the subject of command issuance based on the total power of the power grid, the input demand of each storage and charging system unit, the geographical location, the peak and valley electricity price, and the load forecast value, thereby improving the reliability and flexibility of the system. Through the deep learning algorithm, the scientific nature of the "machine networking" energy scheduling is improved. By dynamically predicting the load of each unit, the system can provide a more accurate basis for competition weights when performing energy scheduling, thereby optimizing the overall energy management strategy and improving the efficiency and stability of the system. This embodiment proposes a "machine networking" structure between decentralized storage and charging system units, which is compatible with the flexible networking mode of 1-n storage and charging units and realizes decentralized control without a host. Through the competition algorithm, each storage and charging unit dynamically decides on the subject of issuing commands based on factors such as the total power input from the power grid, their respective input requirements, geographical location, peak and valley electricity prices, and load forecast values ​​calculated by the deep learning algorithm, thereby optimizing the operation of the entire cluster. In addition, this embodiment also proposes an energy scheduling optimization method based on a deep learning algorithm, which can dynamically predict the load of each unit and provide a basis for competitive weights in the "machine network" energy scheduling process, further improving the energy efficiency and scheduling scientificity of the system. Through the autonomous scheduling mechanism and intelligent algorithm, each storage and charging unit can dynamically optimize the energy distribution on the DC bus according to real-time energy demand and load forecast conditions, avoid the limitations of centralized control methods, and improve the flexibility and stability of the system.

[0071] like Figure 1 As shown in the figure, the EMS (Energy Management System) of the storage and charging unit is built through ZYNQ (FPGA+ARM combined with the main control chip), the hardware-level control response system design is built on the PL (Programmable Logic) side, and the energy management strategy and cloud server interconnection function are built on the PS (Processing System) side. The ARM processing unit provides rich interfaces and powerful computing power, which is suitable for running complex energy management algorithms and exchanging data with the cloud.

[0072] like Figure 2 As shown, a special cable containing two signal lines (S+, S-), two power lines (V+, V-), a ground line (GND), and two thermal liquid tubes is used to connect each storage unit.

[0073] like Figure 3 As shown in the figure, each storage and charging unit is connected in parallel in the machine network by using special cables, and the internal circuit of each storage and charging unit is connected through the machine network connection controller. The nodes in the machine network communicate using the controller area network with flexible rate (CAN-FD) protocol. In this way, the hardware connection and communication control response mechanism of the machine network cluster structure are constructed.

[0074] The system architecture of this implementation combines the powerful functions of the ZYNQ chip, the flexible connection of special cables, and the reliable communication of the CAN-FD protocol to build an efficient, stable, and scalable energy management system (EMS). The system not only realizes the real-time and high efficiency of the hardware-level control response system, but also realizes the intelligentization and remote monitoring of energy management strategies through the interconnection function between the ARM processing unit and the cloud server. This system architecture has broad application prospects in the fields of energy storage and charging.

[0075] like Figure 4 As shown, after each storage and charging unit is connected to the machine network, when a storage and charging unit obtains external power input or needs power assistance, it will actively initiate a competition mechanism in the machine network. After other nodes receive the competition initiation instruction, they will calculate their own competition weight factors and send their own weight information to the machine network within the specified time window. The competition mechanism initiator unit collects the weight data of each unit and determines the external input power distribution object and the assistance object.

[0076] The specific calculation method of competitive value is as follows Figure 5 As shown, each node calculates its own competition value based on the following two factors:

[0077] 1. Remaining power of energy storage (Battery State of Charge, SoC);

[0078] 2. Predicted Load from the Model (PL);

[0079] 3.Price Factor (PF);

[0080] Each factor is added with a certain weight to get the total competitive value of the node. The formula is as follows:

[0081] CV i =w1·SoC i +w2·PL i +w3·PF i

[0082] Among them, CV i is the competition value of node i; SoC i is the remaining energy storage capacity of node i; PLi is the predicted load of node i; PF i is the relative factor of the electricity price level when calculating the competitive value of node i, which is 1 during peak hours, -1 during valley hours, and 0 during normal hours. w1, w2, and w3 are the weights of the corresponding factors, which are determined by regression analysis of historical data to determine the degree of influence of each factor on the system, and are dynamically adjusted through the feedback mechanism during actual operation.

[0083] Among them, PL is the predicted load of the node. It extracts the historical charging data and relative geographical location characteristics of the charging pile by using the variational mode decomposition (VMD) algorithm to construct a data set with spatiotemporal characteristics. The spatiotemporal graph convolutional network (ST-GCN) algorithm is used to build a charging pile charging load prediction model and the gray wolf optimization (GWO) algorithm is used to optimize the model parameters to improve the model performance. The specific implementation steps are as follows: Figure 6 As shown:

[0084] The description of each link is as follows:

[0085] 1. Data preprocessing: Data preprocessing includes two steps: data collection and feature extraction:

[0086] Data collection: Collect historical charging data of each storage and charging unit, including the charging amount, charging time, relative geographical location, etc. in each time period.

[0087] Feature extraction: Extract the time series features and relative geographic position (RGP) features of the charging data of the charging pile, where the RGP feature is the relative geographic position information of a storage unit in the machine network cluster. Assuming there is a storage unit machine network cluster set up in the parking lot of a gas station, the RGP feature value of the storage unit closest to the parking lot entrance should be the highest. The farther the unit is from the entrance, the lower the RGP feature value. This is because the distance from the entrance directly affects the load of the storage unit. The closer the storage unit is to the entrance, the higher the utilization rate.

[0088] 2. Variational Mode Decomposition (VMD):

[0089] Modal decomposition: VMD is performed on the collected historical charging data of charging piles to decompose the original time series into several modal functions. Each modal function represents different frequency components and trends.

[0090] Construction of spatiotemporal feature dataset: The decomposed modal function is combined with the relative geographic location features to form a dataset with spatiotemporal features.

[0091] 3. Build the spatiotemporal graph convolutional network (ST-GCN) model:

[0092] Model architecture design:

[0093] Input layer: Input a data set with spatiotemporal characteristics. The dimension of the input data is (T×N×F), where (T) is the time step, (N) is the number of nodes (i.e. the number of charging piles), and (F) is the feature dimension.

[0094] Temporal convolution layer: One-dimensional convolution (1D-CNN) is used to extract temporal features, and the convolution kernel size is (k t ), the output dimension is (T'×N×F'), where (T') is the length of the time step after convolution and (F') is the output feature dimension.

[0095] Graph convolution layer: Graph convolution (GCN) is used to extract spatial features, and a graph structure is constructed based on the relative geographical locations of each storage unit. The output dimension of the graph convolution layer is (T'×N×F"), where (F") is the feature dimension after graph convolution.

[0096] Temporal convolution layer: One-dimensional convolution is applied again to further extract temporal features, and the output dimension is (T”×N×F”’), where (T”) is the length of the time step after convolution again, and (F”’) is the final feature dimension.

[0097] Fully connected layer: Flatten the features output by the convolutional layer and output the predicted value through the fully connected layer.

[0098] 4. Optimization through the Grey Wolf Algorithm (GWO):

[0099] Objective function definition: The objective function is defined as the mean square error (MSE) between the model prediction result and the actual charging load.

[0100] Parameter optimization: GWO is used to optimize the parameters of the ST-GCN model. Initially, several groups of gray wolves are set up, each individual represents a set of model parameters, and the group position is iteratively updated to find the optimal parameter combination to minimize the objective function value.

[0101] 5. Model training and verification:

[0102] Data segmentation: Divide the constructed spatiotemporal feature dataset into a training set and a validation set.

[0103] Model training: Train the ST-GCN model, use GWO to optimize parameters, and continuously improve model performance through multiple rounds of iterations.

[0104] Model validation: Evaluate model performance on the validation set to ensure that the model has good generalization ability.

[0105] 6. Model deployment and application:

[0106] Model deployment: Deploy the trained ST-GCN model to the charging pile management system to achieve real-time charging load prediction.

[0107] Strategy optimization: Based on the prediction results, optimize the scheduling and management strategies of charging piles to improve charging efficiency and user experience.

[0108] 7. Performance evaluation and improvement:

[0109] Model evaluation: Regularly evaluate the performance of the model in actual applications and collect feedback data for analysis.

[0110] Continuous optimization: Based on the evaluation results, the model is continuously optimized and improved to ensure that it maintains good prediction performance in different environments and conditions.

[0111] Through the above steps, this embodiment can efficiently predict the charging load of the charging pile, provide a scientific basis for the management and scheduling of the charging pile, and improve the utilization rate of the charging pile and user satisfaction.

[0112] The specific application methods of competitive values ​​in two usage scenarios are as follows:

[0113] When the storage unit that obtains external power input initiates an input power allocation request, power allocation is performed according to the competition value of the node. The specific steps are as follows:

[0114] 1. All nodes calculate and report the competitive value CV i ;

[0115] 2. Filter the storage and charging nodes with a competition value less than a certain value to enter the candidate list for input allocation. This is because during the peak electricity price period, the electricity price factor is positive when calculating the competition value, which will increase the competition value of all storage and charging units. During the valley electricity price period, the electricity price factor is negative when calculating the competition value, which will lower the competition value of all storage and charging units. By setting a reasonable competition value threshold for receiving input allocation, it is possible to charge as much as possible during the valley period and as little as possible during the peak period, so as to achieve the purpose of peak shaving and valley filling.

[0116] 3. According to the size of the competition value, the input power P is allocated in proportion. The smaller the competition value, the higher the priority of the storage unit in input allocation. In a possible implementation method, the number of units receiving input (Np) is calculated based on the maximum input power (Pd) acceptable to a single node, that is, Np = P / Pd. The result is rounded to the number of units receiving input, and sorted according to the competition value. The storage unit with the smallest number of units that meets the requirements obtains input allocation;

[0117] 4. After a certain time window (e.g. 10 minutes), the storage unit that receives external power input will initiate a competition mechanism again to re-determine the storage unit that receives input allocation. Since the storage unit with the lowest weight previously received power input, its competition value will increase, causing a ranking change. The storage unit with the lowest new competition value will receive power input.

[0118] When a node needs power assistance, the node that provides power assistance is determined based on the competition value. The specific steps are as follows:

[0119] The unit that needs assistance initiates a request, and all nodes calculate and report the competitive value CV i .

[0120] The unit that needs assistance sorts other units according to the size of the competition value and selects the node with the highest competition value to provide assistance.

[0121] If the power of a node is not enough to meet the demand, multiple nodes can be selected to provide assistance and the assistance power can be allocated according to the proportion of competition value.

[0122] Through the above method, the power competition mechanism can be effectively implemented to ensure that each node competes for power resources reasonably according to its own situation and provides effective power assistance when needed. This not only improves the efficiency of the system, but also enhances the flexibility and reliability of the system.

[0123] The machine networking proposed in this embodiment is applied to the storage and charging unit cluster for performance testing, and the daily usage scenario data (excluding weekends and holidays) during the three-month trial period are collected. The supercharging full service rate and the average ratio of the amount of electricity provided to the electricity cost of the storage and charging unit cluster under each strategy are summarized and calculated to obtain Figure 7 and Figure 8 The data shown. It can be seen from the data that the use of the machine networking method in the embodiment of the present application to build a network can significantly improve the supercharging full service rate, but compared with not using the machine networking, due to the increase in the service rate, the electric energy required from the power grid during the daytime peak period is also significantly increased, and the ratio of the average charging amount to the electricity cost is increased, resulting in an increase in service costs. However, after adopting the neural network optimization strategy, through the flexible regulation of the power grid to obtain electric energy, a supercharging full service rate similar to that of the fixed strategy is achieved, while the ratio of the charging amount to the electricity cost is significantly reduced, and the service cost is significantly reduced.

[0124] An energy management system for a distributed storage and charging system, comprising:

[0125] A machine networking building module, used to build a machine networking architecture including several storage and charging units;

[0126] A competition mechanism module, used for calculating a corresponding competition value based on the energy storage remaining power data and predicted load corresponding to each of the storage and charging units when there is an external power input or power assistance demand in the machine networking architecture;

[0127] The assistance module is used to calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units according to the competition value corresponding to each of the storage and charging units.

[0128] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program so that the electronic device executes the energy management method of a distributed storage and charging system.

[0129] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the energy management method of a distributed storage and charging system.

[0130] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An energy management method for a distributed storage and charging system, characterized in that: include: Step 1: Build a machine network architecture including several storage and charging units; Step 2: When there is an external power input or power assistance demand in the machine networking architecture, a corresponding competition value is calculated based on the energy storage remaining power data and predicted load corresponding to each storage and charging unit; Step 3: Calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units based on the competition value corresponding to each of the storage and charging units.

2. The energy management method of a distributed storage and charging system according to claim 1, characterized in that: The step 2 specifically includes: Acquire training data, wherein the training data includes historical charging data, relative geographical location data and load data of each of the storage and charging units; Construct a load forecasting model based on spatiotemporal graph convolutional network; Based on the training data and in combination with the Grey Wolf Optimization Algorithm, the load forecasting model is trained and parameter optimization is performed to obtain an optimized load forecasting model; Execute load prediction of each storage unit based on the optimized load prediction model to obtain predicted load data of each storage unit; Obtaining the remaining energy storage capacity data and electricity price level data of each of the storage and charging units; The energy storage remaining capacity data, electricity price level data and predicted load data of each storage and charging unit are weighted and summed to obtain a corresponding competitive value.

3. The energy management method of a distributed storage and charging system according to claim 2, characterized in that: The load forecasting model is trained and parameter optimization is performed based on the training data and combined with the Grey Wolf Optimization Algorithm, specifically including: Extracting a hyperparameter combination included in the load forecasting model; Based on the training data, the load forecasting model corresponding to each of the hyperparameter combinations is trained, and the training results are evaluated, and the evaluation results are used as the evaluation scores corresponding to each of the hyperparameter combinations; Generating a gray wolf population for gray wolves using each of the hyperparameter combinations; Based on the evaluation scores corresponding to the hyperparameter combinations and in combination with the gray wolf optimization algorithm, parameter optimization is performed within the gray wolf population to obtain the optimal hyperparameter combination; An optimized load forecasting model is constructed based on the optimal hyperparameter combination.

4. The energy management method of a distributed storage and charging system according to claim 3 is characterized in that: Training the load forecasting model corresponding to each of the hyperparameter combinations based on the training data specifically includes: Dividing the training data into a training set, a test set, and a validation set according to a preset ratio; The load prediction model corresponding to each of the hyperparameter combinations is trained based on the training set, and the performance of the load prediction model corresponding to each of the hyperparameter combinations is verified and evaluated through the verification set, and the evaluation result is used as the evaluation score corresponding to each of the hyperparameter combinations.

5. The energy management method of a distributed storage and charging system according to claim 3 is characterized in that: Based on the evaluation scores corresponding to the hyperparameter combinations and in combination with the gray wolf optimization algorithm, parameter optimization is performed within the gray wolf population, specifically including: Setting the value range, population size and number of iterations of the hyperparameter combination, and using each of the hyperparameter combinations as gray wolves to generate the gray wolf population; Based on the gray wolf optimization algorithm, each hyperparameter combination in the gray wolf population is iteratively updated multiple times. In each iterative update process, the evaluation score corresponding to each hyperparameter combination is calculated, and the gray wolf population is updated based on the evaluation score corresponding to each hyperparameter combination until the number of iterations reaches a preset value and the optimal hyperparameter combination is output.

6. The energy management method of a distributed storage and charging system according to claim 1, characterized in that: The load prediction of each storage unit is performed based on the optimized load prediction model, specifically including: Extracting the time series characteristics and relative geographical location characteristics of the storage unit, Performing modal decomposition on the time series features, combining the decomposed modal functions with relative geographical location features, and constructing a feature data set with spatiotemporal features; The load forecasting model includes a first time convolution layer, a graph convolution layer, a second time convolution layer and a fully connected prediction layer connected in sequence, the feature data set is input into the load forecasting model, and features are extracted through each time convolution layer and graph convolution layer; The extracted features are input into the fully connected prediction layer for classification prediction to obtain predicted load data.

7. The energy management method of a distributed storage and charging system according to claim 1, characterized in that: The external electric energy input distribution scheme and the electric energy assistance scheme between the storage and charging units are calculated based on the competitive values ​​corresponding to the storage and charging units, specifically including: Input allocation scheme of external electric energy: sorting the competition values ​​corresponding to the storage and charging units, and selecting a number of storage and charging units as candidate storage and charging units in the sorting based on a preset competition threshold; Based on the competition value corresponding to each candidate storage and charging unit, the input power of external electric energy is allocated in proportion to complete the external electric energy input allocation of the current cycle; Electric energy assistance scheme between storage and charging units: the storage and charging unit to be assisted initiates an assistance request, and sorts the remaining storage and charging units except the storage and charging unit to be assisted based on the competition value, and selects the assisting storage and charging unit for assistance based on the sorting result.

8. An energy management system for a distributed storage and charging system, characterized in that: include: A machine networking building module, used to build a machine networking architecture including several storage and charging units; A competition mechanism module, used for calculating a corresponding competition value based on the energy storage remaining power data and predicted load corresponding to each of the storage and charging units when there is an external power input or power assistance demand in the machine networking architecture; The assistance module is used to calculate the input distribution plan of external electric energy and the electric energy assistance plan between the storage and charging units according to the competition value corresponding to each of the storage and charging units.

9. An electronic device, characterized in that: It includes a memory and a processor, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute an energy management method for a distributed storage and charging system according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements an energy management method for a distributed storage and charging system as described in any one of claims 1-7.

Images