Electric vehicle charging station load prediction method, device and equipment based on graph neural network
Through the method based on graph neural network, the load prediction of electric vehicle charging stations is solved, and the stable operation of the power grid and the utilization rate of charging facilities is improved.
Patent Information
- Application Number
- CN202411148637.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Large-scale electric vehicles are connected to the power grid and cause changes in the power grid volatility, affecting the reliability, safety and economics of the power grid.
Using a graph-based neural network method, by obtaining the historical load data of electric vehicle charging stations, combining the road network distance data and charging station power consumption data, the sequence and relationship embedding variables are extracted using LSTM and TGC networks to perform full connection layer prediction, so as to Achieve accurate prediction of the load of multiple charging stations.
The grid load operation curve has been optimized, the grid is operated stably, the utilization rate of charging facilities has been improved, and the negative impact of electric vehicle charging on the grid has been reduced.
Smart Images

Figure CN118941050A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of electric vehicles and load forecasting technology, and in particular to a method, device and equipment for forecasting load at an electric vehicle charging station based on a graph neural network. Background Art
[0002] As the problem of environmental degradation becomes increasingly prominent, and due to the shortage of resources and pollution problems of traditional non-renewable energy sources such as coal and oil, energy consumption is transforming towards electrification, cleanliness, low carbonization and intelligence. Electric vehicles are gradually gaining attention due to their advantages such as high efficiency, zero pollution and low emissions. The electric vehicle industry is developing rapidly, and large-scale electric vehicles are connected to the power grid, causing fluctuations in the power grid, affecting the operation of the power grid in terms of reliability, safety and economy. It is very necessary to predict its charging load. Accurate prediction helps promote the interaction between electric vehicles and the power grid, realize orderly charging, and then optimize the load operation curve of the power grid, ensure the stable operation of the power grid, improve the utilization rate of charging facilities, and reduce the negative impact of electric vehicle charging on the power grid. Summary of the invention
[0003] The purpose of the present invention is to provide a method, device and equipment for predicting the load of electric vehicle charging stations based on graph neural networks, aiming to solve the problem that large-scale electric vehicles connected to the power grid cause fluctuations in the power grid, affecting the operation of the power grid in terms of reliability, safety and economy.
[0004] According to one aspect of the present disclosure, a method for predicting load of an electric vehicle charging station based on a graph neural network is provided, comprising: Obtain historical load data of N electric vehicle charging stations in each area of the city to be predicted, and determine a road network distance dataset M1 and a charging station power consumption dataset M2 according to the historical load data; Input the charging station power consumption data set M2 into the LSMT network to obtain a sequence embedding variable Et; Input the sequence embedding variable Et and the road network distance data set M1 into the TGC network to obtain a relational embedding variable; The sequence embedding variable Et and the relation embedding variable are input into the fully connected layer to predict the load of the charging station at any time.
[0005] In some implementations of the first aspect, the historical load data includes N active power data, N charging station capacity data, and N charging station road network distance data.
[0006] In some implementations of the first aspect, determining the road network distance dataset M1 and the charging station power consumption dataset M2 according to the historical load data includes: The N active power data are used as a power data set A1, the N charging station capacity data are used as a capacity data set B1, and the N charging station road network distance data are used as a road network distance data set M1; The charging station power consumption dataset M2 is determined according to the power dataset A1 and the capacity dataset B1.
[0007] In some implementations of the first aspect, the charging station power consumption dataset M2 is input into the LSMT network to obtain a sequence embedding variable E t ,include: According to the charging station power consumption data set M2, the charging station power consumption data of any time step is input into the LSMT network; Use the LSMT network to process the charging station power consumption data of any time step and extract the last hidden state ; The last hidden state As the sequence embedding vector of charging station load ; Embedding vector according to the sequence , calculate the sequence embedding variable E t .
[0008] In some implementations of the first aspect, the embedding according to the sequence , calculate the sequence embedding variable E t The calculation formula is: ; in, Indicates at a point in time , the number of charging stations processed by the LSTM layer, U is the number of hidden units in the LSTM, Refers to a real number Dimensional matrix.
[0009] In some implementations of the first aspect, the sequence is embedded in the variable E t And the road network distance dataset M1 is input into the TGC network to obtain the relational embedding variable ,include: Constructing an adjacency matrix between charging stations according to the road network distance dataset M1; Embed the sequence into variable E t and adjacency matrix are input into the TGC network to determine the relationship mapping function; Calculate the relationship embedding variable according to the sequence embedding vector, the relationship mapping function and the road network distance dataset M1 .
[0010] In some implementations of the first aspect, the relationship embedding variable The calculation formula is: ; in, represents the road network distance from charging station i to charging station j, is a relational mapping function, is the adjacency matrix of charging station i and charging station j.
[0011] In some implementations of the first aspect, the relationship mapping function The calculation formula is: ; in, It means concatenating three transposed matrices. and b are neural network parameters; It is represented as the activation function.
[0012] According to a second aspect of the present disclosure, a graph neural network-based electric vehicle charging station load prediction device is provided, comprising: A data acquisition module is used to acquire historical load data of N electric vehicle charging stations in various areas of the city to be predicted, and determine a road network distance dataset M1 and a charging station power consumption dataset M2 according to the historical load data; A sequence embedding module, used to input the charging station power consumption data set M2 into the LSMT network to obtain a sequence embedding variable Et; The relation embedding module is used to input the sequence embedding variable Et and the road network distance dataset M1 into the TGC network to obtain the relation embedding variable ; Load forecasting module, used to embed the sequence into variable Et and the relationship into variable Input the fully connected layer to predict the load of the charging station at any time.
[0013] According to a third aspect of the present disclosure, an electronic device is provided, comprising a processor, and a memory and a network interface connected to the processor; the network interface is connected to a non-volatile memory in a server; when the processor is running, it retrieves a computer program from the non-volatile memory through the network interface, and runs the computer program through the memory to execute the electric vehicle charging station load prediction method based on graph neural network as described above.
[0014] According to a fourth aspect of the present disclosure, a storage medium is provided, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the electric vehicle charging station load prediction method based on graph neural network as described above is implemented.
[0015] The present invention proposes a method for load forecasting that considers both load space factors and external factors, and can coordinate multiple charging stations. First, the historical time series data of each charging station is input into a long short-term memory (LSTM) network to capture sequence dependencies and learn a charging station sequence embedding. Then, by designing a new time series graph convolution, the charging station is regarded as a charging load node, and a load weighted directed graph of all charging stations at each moment is constructed to characterize the spatial relationship between the charging stations. The directed graph is stacked and input into a spatiotemporal neural network to extract spatiotemporal factor features, and the LSTM network is used to extract the features of external factors such as date and historical load. Finally, the cascade of sequence embedding and relationship embedding is fed to a fully connected neural network layer to achieve accurate load forecasting for multiple charging stations, which helps to promote the interaction between electric vehicles and power grids, realize orderly charging, and then optimize the load operation curve of the power grid, ensure the stable operation of the power grid, improve the utilization rate of charging facilities, and reduce the negative impact of electric vehicle charging on the power grid. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A flow chart of a method for predicting load at an electric vehicle charging station based on a graph neural network according to a first embodiment of the present invention is shown; Figure 2 The TCS framework diagram of the first embodiment of the present invention is shown; Figure 3 The accuracy of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network of the second embodiment of the present invention is shown; Figure 4 The accuracy of the conventional method of the second embodiment of the present invention is shown; Figure 5 The mean absolute error of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network according to the second embodiment of the present invention is shown; Figure 6 The mean absolute error of the conventional method of the second embodiment of the present invention is shown; Figure 7 The root mean square error of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network according to the second embodiment of the present invention is shown; Figure 8 The root mean square error of the conventional method of the second embodiment of the present invention is shown; Fig. 9 A block diagram of a distributed smart grid energy system dispatching device based on reward function evaluation according to a fourth embodiment of the present invention is shown; Fig.10 A schematic diagram of a computing device according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0019] The present invention provides a method for predicting the load of an electric vehicle charging station based on a graph neural network, which specifically includes: obtaining historical load data of N electric vehicle charging stations in various areas of a city to be predicted, determining a road network distance data set M1 and a charging station power consumption data set M2 according to the historical load data; inputting the charging station power consumption data set M2 into an LSMT network to obtain a sequence embedding variable Et; embedding the sequence into a variable Et. t and the road network distance dataset M1 are input into the TGC network to obtain a relation embedding variable; the sequence is embedded into the variable E t The relational embedding variables are input into the fully connected layer to predict the load of the charging station at any time. In this way, accurate load prediction for multiple charging stations can be achieved.
[0020] The following is combined with Figure 1-2 The specific embodiments provide a more specific explanation of the electric vehicle charging station load prediction method based on graph neural network provided by the embodiments of the present disclosure.
[0021] Figure 1 A flow chart of a method for predicting electric vehicle charging station load based on a graph neural network provided by an embodiment of the present disclosure is shown. Figure 1 As shown, the method 100 may include the following steps: S101, obtaining historical load data of N electric vehicle charging stations in various areas of a city to be predicted, and determining a road network distance dataset M1 and a charging station power consumption dataset M2 according to the historical load data.
[0022] Among them, the historical load data includes N active power data, N charging station capacity data and N charging station road network distance data.
[0023] In some embodiments, the historical load data of electric vehicle charging stations in various areas within the city to be predicted is obtained, and the implementation process is as follows: Assume that the extraction time range is one week and the extraction interval is 15 minutes, and extract the historical data of the power grid in a certain area within one week, wherein the historical data of the power grid includes active power P, charging station capacity C, and charging station road network distance L; the extracted N active power data are grouped into a power data set A1, the extracted N charging station capacity data are grouped into a capacity data set B1, and the extracted N charging station road network distance data are grouped into a road network distance data set M1; The power consumed by the charging station is recorded as SOC, and the SOC is expressed as:
[0024] According to the above SOC calculation formula, combined with the power data set A1 and the capacity data set B1, the SOC data set can be obtained. The power consumed by the charging station is recorded as the SOC data set as M2. Furthermore, the datasets M1 and M2 are normalized to , the normalized expression is as follows:
[0025] in, is the minimum value of the data set, is the maximum value of the data set.
[0026] S102, input the charging station power consumption data set M2 into the LSMT network to obtain a sequence embedding variable E t .
[0027] In some embodiments, Figure 2 As shown, the present invention proposes a time-varying charging station (TCS) framework to solve the problem of electric vehicle charging station load prediction. The framework includes three layers, namely, sequence embedding layer, relation embedding layer and prediction layer.
[0028] The sequence embedding layer is used to input the charging station power consumption dataset M2 into the LSMT network to obtain the sequence embedding variable E t , specifically, Let t be any time step, Represents an input vector, where D is the input dimension. and Denote the memory state vector and the hidden state vector respectively, where U is the number of hidden units. It is an information conversion module. , and Denote input, output and forget gate respectively. Formally, the transformation module, state vector and control gate are defined by the following formula:
[0029]
[0030]
[0031] Where: , , , , , , is the mapping matrix; , , and is the bias vector. The update formula can be understood as performing the following steps: (1) Calculate from input Information about transitions to memory states By updating ; (2) Update the input gate ,control arrive information; (3) Update the forget gate to decide how much information should be retained in the memory state; (4) Refresh the memory state by fusing the information flow of the input gate and the memory gate ; (5) Update the output gate , adjust the amount of information that can be output; (6) Update hidden state .
[0032] If several charging stations are in the same area, their charging load relationships have similar trends. The historical time series data is input into an LSTM network and the last hidden state As charging station load (note ) ,remember The sequence embedding variable output by LSTM is expressed as follows: (2) Among them, U represents the number of hidden units in LSTM.
[0033] S103, embedding the sequence into a variable E t And the road network distance dataset M1 is input into the TGC network to obtain the relational embedding variable .
[0034] In some embodiments, the sequence can be embedded in the variable The dataset M1 is input into a temporal graph convolution (TGC) neural network to capture the time-varying spatial dependencies and obtain the relational embedding variables .
[0035] Specifically, we first construct the spatiotemporal information graph of charging stations and define the directed graph ,in, for A collection of charging station nodes, For N charging stations The charging load value at the moment, For the Charging stations at all times The charging load value; is the set of edges; is the adjacency matrix, which is a matrix with dimension Based on the construction of the directed graph, the charging station load forecasting problem can be converted into a matrix containing Charging stations Directed graph of historical load values at each moment to predict the future The charging station load in the entire area during the time period is expressed as: (4) Where: are neural network parameters, including edge sets, adjacency matrices, etc.; For prediction The collection of load forecast data of all charging stations at all times; for The load of all charging stations observed at all times.
[0036] Secondly, the sequence embedding variable Et and the adjacency matrix are input into the TGC network to determine the relationship mapping function , according to the sequence embedding vector, relationship mapping function and the road network distance dataset M1, calculate the relationship embedding variable When two charging stations CSi and CSj affect each other, the relationship between them is likely to affect their loads in the near future. The non-uniformity coefficient of the relationship between charging station i and charging station j is , which is expressed as follows:
[0037] in, represents the road network distance from charging station i to charging station j, is a relational mapping function, is the adjacency matrix of charging station i and charging station j, and the relationship mapping function It is expressed as follows:
[0038] in, It means concatenating three transposed matrices. and b are neural network parameters; It is represented as the activation function.
[0039] S104, embedding the sequence into a variable E t and relational embedding variables Input the fully connected layer to predict the load of the charging station at any time.
[0040] In some embodiments, the sequence embedding variable Et and the relation embedding variable Enter a fully connected layer to achieve accurate load forecasting for multiple charging stations.
[0041] According to the embodiments of the present invention, the following technical effects are achieved: The present invention proposes a method for load forecasting that considers both load space factors and external factors, and can coordinate multiple charging stations. First, the historical time series data of each charging station is input into a long short-term memory (LSTM) network to capture sequence dependencies and learn a charging station sequence embedding. Then, by designing a new time series graph convolution, the charging station is regarded as a charging load node, and a load weighted directed graph of all charging stations at each moment is constructed to characterize the spatial relationship between the charging stations. The directed graph is stacked and input into a spatiotemporal neural network to extract spatiotemporal factor features, and the LSTM network is used to extract the features of external factors such as date and historical load. Finally, the cascade of sequence embedding and relationship embedding is fed to a fully connected neural network layer to achieve accurate load forecasting for multiple charging stations, which helps to promote the interaction between electric vehicles and power grids, realize orderly charging, and then optimize the load operation curve of the power grid, ensure the stable operation of the power grid, improve the utilization rate of charging facilities, and reduce the negative impact of electric vehicle charging on the power grid.
[0042] Embodiment 2 The following is combined with Figure 3-8 The specific implementation example provides a more specific description of the electric vehicle charging station load prediction method based on graph neural network.
[0043] Obtain the historical load data of electric vehicle charging stations in various areas within the city to be predicted. The implementation process is as follows: Assume that the number of extracted charging stations is 98, the time range is one week, and the extraction interval is 15 minutes. The historical data of the power grid in a certain area within one week are extracted. The historical data of the power grid include active power P, charging station capacity C, and charging station road network distance L. The extracted 24192 active power data are combined into a power data set A1, the extracted 24192 charging station capacity data are combined into a capacity data set B1, and the extracted 24192 charging station road network distance data are combined into a road network distance data set M1. Using Python language and based on the Tensorflow deep learning framework, the graph neural network electric vehicle charging station load prediction method based on time-varying relationship of the present invention is implemented. The prediction results are as follows Figure 3-8 shown.
[0044] like Figure 3 and Figure 4 As shown, it can be found from the experimental results that the accuracy of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network of the present invention is about 91%, while the accuracy of the traditional method can only reach 85% at most under equal conditions.
[0045] like Figure 5 and Figure 6As shown, from the experimental results, it can be found that the average absolute error of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network of the present invention can be reduced to less than 5%, while the accuracy of the traditional method can only be reduced to 8% at most under equal conditions.
[0046] like Figure 7 and Figure 8 As shown, it can be found from the experimental results that the root mean square error of the electric vehicle charging station load prediction method based on the time-varying relationship graph neural network of the present invention is also significantly lower than that of the traditional method, and the error is more stable.
[0047] In summary, the present invention realizes accurate load prediction for multiple charging stations, which helps to promote the interaction between electric vehicles and the power grid, realize orderly charging, and further optimize the load operation curve of the power grid, ensure the stable operation of the power grid, improve the utilization rate of charging facilities, and reduce the negative impact of electric vehicle charging on the power grid.
[0048] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present invention.
[0049] The above is an introduction to a method embodiment. The following is a further explanation of the solution of the present invention through an apparatus embodiment.
[0050] Embodiment 3 Fig. 9 A block diagram of a graph neural network-based electric vehicle charging station load prediction device 900 according to an embodiment of the present disclosure is shown. Fig. 9 As shown, the device 900 includes: The data acquisition module 910 is used to acquire the historical load data of N electric vehicle charging stations in each area of the city to be predicted, and determine the road network distance data set M1 and the charging station power consumption data set M2 according to the historical load data; The sequence embedding module 920 is used to input the charging station power consumption data set M2 into the LSMT network to obtain a sequence embedding variable E t ; The relation embedding module 930 is used to embed the sequence into a variable E t and the road network distance dataset M1 are input into the TGC network to obtain relational embedding variables; The load forecasting module 940 is used to embed the sequence into the variable E tThe relationship embedding variables are input into the fully connected layer to predict the load of the charging station at any time. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.
[0051] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.
[0052] Embodiment 4 Fig.10 A schematic structural diagram of an electronic device 500 that can be used to implement an embodiment of the present invention is shown.
[0053] like Fig.10 As shown, the electronic device 110 includes at least one processor 111, and a memory connected to the at least one processor 111 in communication, such as a read-only memory (ROM) 112, a random access memory (RAM) 113, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 111 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 112 or the computer program loaded from the storage unit 118 to the random access memory (RAM) 113. In the RAM 113, various programs and data required for the operation of the electronic device 100 can also be stored. The processor 111, the ROM 112, and the RAM 113 are connected to each other via a bus 114. An input / output (I / O) interface 115 is also connected to the bus 114.
[0054] A number of components in the electronic device 110 are connected to the I / O interface 115, including: an input unit 116, such as a keyboard, a mouse, etc.; an output unit 117, such as various types of displays, speakers, etc.; a storage unit 118, such as a disk, an optical disk, etc.; and a communication unit 119, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 119 allows the electronic device 110 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0055] The processor 111 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 111 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 111 executes the various methods and processes described above, such as a distributed smart grid energy system scheduling method based on reward function evaluation.
[0056] In some embodiments, the distributed smart grid energy system scheduling method based on reward function evaluation can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 118. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 110 via the ROM 112 and / or the communication unit 119. When the computer program is loaded into the RAM 113 and executed by the processor 111, one or more steps of the angle adjustment method of the leveling cylinder and the luffing cylinder described above can be executed. Alternatively, in other embodiments, the processor 111 can be configured to execute the distributed smart grid energy system scheduling method based on reward function evaluation in any other appropriate manner (for example, by means of firmware).
[0057] Various implementations 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 chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0058] Computer programs for implementing 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, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.
[0059] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0060] To provide interaction with a user, the systems and techniques described herein may 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 a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).
[0061] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0062] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.
[0063] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document is not limited here. The above is only a specific implementation mode of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.
Claims
1. The electric vehicle charging station load forecasting method based on graph neural network is characterized by: include: Obtain historical load data of N electric vehicle charging stations in each area of the city to be predicted, and determine a road network distance dataset M1 and a charging station power consumption dataset M2 according to the historical load data; The charging station power consumption dataset M2 is input into the LSMT network to obtain the sequence embedding variable E t ; Embed the sequence into variable E t And the road network distance dataset M1 is input into the TGC network to obtain the relational embedding variable ; Embed the sequence into variable E t and relational embedding variables Input the fully connected layer to predict the load of the charging station at any time.
2. The method according to claim 1, characterized in that The historical load data includes N active power data, N charging station capacity data and N charging station road network distance data.
3. The method according to claim 2, characterized in that The determining of the road network distance dataset M1 and the charging station power consumption dataset M2 according to the historical load data includes: The N active power data are used as a power data set A1, the N charging station capacity data are used as a capacity data set B1, and the N charging station road network distance data are used as a road network distance data set M1; The charging station power consumption dataset M2 is determined according to the power dataset A1 and the capacity dataset B1.
4. The method according to claim 1, characterized in that The charging station power consumption dataset M2 is input into the LSMT network to obtain the sequence embedding variable E t ,include: According to the charging station power consumption data set M2, the charging station power consumption data of any time step is input into the LSMT network; Use the LSMT network to process the charging station power consumption data of any time step and extract the last hidden state ; The last hidden state As the sequence embedding vector of charging station load ; Embedding vectors according to the sequence , calculate the sequence embedding variable .
5. The method according to claim 4, characterized in that The embedding according to the sequence , calculate the sequence embedding variable E t The calculation formula is: in, Indicates at a point in time , the number of charging stations processed by the LSTM layer, U is the number of hidden units in the LSTM, Refers to a real number Dimensional matrix.
6. The method according to any one of claims 1 or 4, characterized in that: Embed the sequence into variable E t And the road network distance dataset M1 is input into the TGC network to obtain the relational embedding variable ,include: Constructing an adjacency matrix between charging stations according to the road network distance dataset M1; Embed the sequence into variable E t and adjacency matrix are input into the TGC network to determine the relationship mapping function; Calculate the relationship embedding variable according to the sequence embedding vector, the relationship mapping function and the road network distance dataset M1 .
7. The method according to claim 6, characterized in that The relationship embeds the variable The calculation formula is: in, represents the road network distance from charging station i to charging station j, is a relational mapping function, is the adjacency matrix of charging station i and charging station j.
8. The method according to claim 7, characterized in that The relationship mapping function The calculation formula is: in, It means concatenating three transposed matrices. and b are neural network parameters; It is represented as the activation function.
9. The electric vehicle charging station load prediction device based on graph neural network is characterized by: include: A data acquisition module is used to acquire historical load data of N electric vehicle charging stations in various areas of the city to be predicted, and determine a road network distance dataset M1 and a charging station power consumption dataset M2 according to the historical load data; The sequence embedding module is used to input the charging station power consumption data set M2 into the LSMT network to obtain the sequence embedding variable E t ; The relation embedding module is used to embed the sequence into the variable E t And the road network distance dataset M1 is input into the TGC network to obtain the relational embedding variable ; Load forecasting module, used to embed the sequence into variable E t and relational embedding variables Input the fully connected layer to predict the load of the charging station at any time.
10. An electronic device, characterized in that: It includes a processor, and a memory and a network interface connected to the processor; the network interface is connected to a non-volatile memory in a server; when the processor is running, it calls a computer program from the non-volatile memory through the network interface, and runs the computer program through the memory to execute the method described in any one of claims 1 to 8.
Citation Information
Patent Citations
Method for predicting charging load of electric vehicle
CN113627661A
Electric vehicle charging load space-time distribution prediction method based on graph neural network
CN114358449A
Regional electric vehicle charging load prediction method
CN115907122A
Electric vehicle charging station load prediction method and system based on multi-source feature fusion
CN118487264A
Charging station cluster load prediction method and system based on deep fusion of exogenous variables
CN118504792A
Cited By
Processing method for predicting node load based on electric vehicle data
CN119721383A
A processing method for predicting node load based on electric vehicle data
CN119721383B
