A method and system for virtual collection of regional electric vehicle charging load

Through dynamic time regularization and CNN-GRU neural network model, combined with hierarchical clustering and smoothing processing, virtual acquisition of electric vehicle loads is realized, solving the problem of real-time acquisition of electric vehicle loads is reduced, and the acquisition cost is improved, and the grid management efficiency is improved.

CN118117757BActive Publication Date: 2024-11-12ZHEJIANG UNIV +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410433130.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-11
Publication Date
2024-11-12
Estimated Expiration
2044-04-11

AI Technical Summary

Technical Problem

The prior art cannot collect real-time power data of regional-level electric vehicle loads in real time, and traditional load predictions are difficult to accurately evaluate electric vehicle loads, and consume a lot of hardware acquisition and communication costs.

Method used

The dynamic time regularization algorithm is used to screen the benchmark table area, and the CNN-GRU hybrid neural network model training is combined with hierarchical clustering and smoothing processing to realize virtual acquisition of electric vehicle loads in the selected area.

Benefits of technology

In real-time state, only some station area data need to be collected, which can accurately estimate the load of electric vehicles in a wide range of areas, improving the efficiency and reliability of power grid operation and maintenance management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118117757B_ABST
    Figure CN118117757B_ABST
Patent Text Reader

Abstract

The present invention discloses a virtual collection method and system for regional electric vehicle charging load, including: obtaining the historical data set of electric vehicle load of each substation and the historical data of the total electric vehicle load of the region; using a dynamic time warping algorithm to capture the similarity of the time series of historical data of each substation and the total region, and screening the benchmark substation; calculating the similarity matrix and clustering and reducing the dimension, smoothing the data, and obtaining the input of the neural network model; building a CNN-GRU hybrid neural network model, using the historical data as a training set for training, and collecting the electric vehicle meter data of the selected benchmark substation in real time, using the clustering and smoothing results as the input of the virtual collection model, and obtaining the real-time value of the regional electric vehicle load. The present invention only needs to collect the electric vehicle meter data of the benchmark substation in real time, and can perform real-time virtual collection of the electric vehicle load data in the entire region, reducing the cost of communication transmission.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicles, and in particular to a method and system for virtual collection of charging loads of regional electric vehicles. Background Art

[0002] Electric vehicle loads are usually coupled with other power loads in the substation, and the real-time power value of electric vehicle loads cannot be directly monitored through the substation transformer. If you want to collect real-time power data of regional electric vehicle loads in real time, you need to monitor all smart meters in each substation, which requires a lot of hardware collection and communication costs. The charging behavior of electric vehicle users is highly uncertain and is affected by many factors. It is difficult to accurately evaluate the electric vehicle load in real time using traditional day-ahead load forecasting technology. Summary of the invention

[0003] The purpose of the present invention is to propose a regional electric vehicle charging load virtual collection method to address the deficiencies in the prior art, so as to solve the problem that the existing communication link cannot perform real-time collection of a wide range of massive electric vehicle loads. The present invention can virtually collect the total load power of electric vehicles in all substations under the selected area by only collecting the electric vehicle load power data under some substations in real time.

[0004] The object of the present invention is achieved through the following technical solution: A method for virtual collection of regional electric vehicle charging loads, comprising the following steps:

[0005] S1. Obtain the historical data set of electric vehicle load in each substation and the historical data of the total electric vehicle load in the region;

[0006] S2. Use dynamic time warping algorithm to capture the similarity of the time series of the historical data of electric vehicle load in each area and the historical data of the total electric vehicle load in the region; use the time series similarity to screen the benchmark area;

[0007] S3, calculating the similarity matrix of the time series similarity of each benchmark station area, clustering and reducing the dimension of the similarity matrix using a hierarchical clustering method, smoothing various types of data, and obtaining the input of the neural network model;

[0008] S4. Build a CNN-GRU hybrid neural network model, use the features and historical data of regional total electric vehicle load as training sets, train the neural network model, and obtain a regional electric vehicle charging load virtual collection model;

[0009] S5. Collect electric vehicle meter data of the selected benchmark substation in real time, and use the hierarchical clustering method and smoothing processing results as input of the regional electric vehicle charging load virtual collection model to obtain the real-time value of the regional electric vehicle load.

[0010] Furthermore, S1 specifically includes: linearly interpolating and filling missing values ​​of the historical data of electric vehicle load recorded by each meter in the selected area, accumulating the power information recorded by the electric vehicle meters in each substation to obtain the historical data of the electric vehicle load in each substation, and accumulating and summing the historical data of the electric vehicle load in each substation to obtain the historical data of the total electric vehicle load in the region.

[0011] Furthermore, the dynamic time warping algorithm is used to capture the similarity between the historical data of electric vehicle load in each substation and the historical data time series of the regional total electric vehicle load. Specifically, the dynamic time warping algorithm is used to calculate the similarity, obtain the shortest distance between the sequences, sort the shortest distances, and select several substations with the shortest distance from the regional total electric vehicle load data sequence as benchmark substations. The formula is as follows:

[0012]

[0013] in, For Taiwan Series data of electric vehicle load power; is the series data of regional total electric vehicle load power; Indicates that the dynamic time warping algorithm is used to calculate the similarity of two sequence data.

[0014] Furthermore, in the similarity calculation, the continuous historical load data set is divided into multiple independent data subsets, and similarity analysis is performed independently on each data subset to calculate the similarity index within it; finally, the similarity indexes of all subsets are accumulated and summed to obtain the total similarity of the entire data set.

[0015] Furthermore, the smoothing process for each type of data is specifically as follows: according to the result of similarity matrix clustering dimensionality reduction, the historical load data of electric vehicle loads in each substation in each type obtained by clustering are added together.

[0016] Furthermore, the CNN-GRU hybrid neural network model specifically includes: using the convolution layer of CNN to extract the characteristics and patterns of the smoothed electric vehicle historical load data, and using the pooling layer of CNN to achieve output dimensionality reduction of the convolution layer, and then using the GRU layer to learn the time series characteristics of electric vehicle load power changes. Its input data passes through the input layer, convolution layer, pooling layer, GRU layer, and fully connected layer, and then outputs the regional electric vehicle charging load virtual collection results through the output layer.

[0017] Furthermore, a neuron output discarding index is set in the GRU layer, and when the GRU layer performs output, output discarding is performed according to the neuron output discarding index to prevent overfitting.

[0018] According to a second aspect of the present specification, there is provided a regional electric vehicle charging load virtual collection system, the system comprising a load collection module, a benchmark station area screening module, a data processing module and a virtual collection module;

[0019] The load acquisition module is used to obtain the historical data set of electric vehicle load in each substation and the historical data of the total electric vehicle load in the region, and subsequently collect the electric vehicle meter data of the selected benchmark substation in real time;

[0020] The benchmark area screening module uses a dynamic time warping algorithm to capture the similarity of the time series of the historical data of the electric vehicle load in each area and the historical data of the total electric vehicle load in the region; and uses the time series similarity to screen the benchmark area;

[0021] The data processing module is used to calculate the similarity matrix of the time series similarity of each benchmark station area, cluster and reduce the dimension of the similarity matrix using a hierarchical clustering method, and perform smoothing on various types of data.

[0022] The virtual acquisition module is used to build a CNN-GRU hybrid neural network model. The features and historical data of the total electric vehicle load in the region are used as training sets to train the neural network model to obtain a regional electric vehicle charging load virtual acquisition model. Clustering and smoothing processing are performed based on the real-time collected electric vehicle meter data of the selected benchmark substation. The results are used as the input of the regional electric vehicle charging load virtual acquisition model to obtain the real-time value of the regional electric vehicle load.

[0023] According to a third aspect of the present specification, a regional electric vehicle charging load virtual collection device is provided, comprising a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, a regional electric vehicle charging load virtual collection method is implemented.

[0024] According to a fourth aspect of the present specification, there is provided a computer-readable storage medium having a program stored thereon, wherein when the program is executed by a processor, a method for virtual collection of regional electric vehicle charging load is implemented.

[0025] Beneficial effects of the present invention:

[0026] By using the regional electric vehicle charging load virtual collection model proposed in the present invention, only a small part of the electric vehicle load data under the selected substation needs to be collected in real time, and no other additional information support is required, so that accurate estimation of a large number of electric vehicle loads in a wide area can be achieved. The present invention provides an overall framework for implementing a virtual collection method in the electric vehicle field. The technology can track the changes in a wide range of massive electric vehicle loads in real time with high accuracy, which is of great significance for the short-term scheduling of electric vehicle power grids and helps to improve the efficiency and reliability of power grid operation and maintenance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a method for virtual collection of regional electric vehicle charging loads provided by an embodiment of the present invention;

[0028] Figure 2 It is a framework diagram of a method for virtual collection of regional electric vehicle charging loads provided by an embodiment of the present invention;

[0029] Figure 3 It is a CNN-GRU hybrid neural network result diagram provided by an embodiment of the present invention;

[0030] Figure 4 It is a schematic diagram of the hierarchical analysis method clustering of the selected 100 benchmark stations provided in an embodiment of the present invention;

[0031] Figure 5 This is a diagram showing the comparison results of different schemes provided by the embodiments of the present invention;

[0032] Figure 6 Schematic diagram of the comparison between the virtual collection curve of the regional electric vehicle total load provided by the embodiment of the present invention and the actual regional electric vehicle total load curve Figure 1 ;

[0033] Figure 7 Schematic diagram of the comparison between the virtual collection curve of the regional electric vehicle total load provided by the embodiment of the present invention and the actual regional electric vehicle total load curve Figure 2 ;

[0034] Figure 8 Schematic diagram of the comparison between the virtual collection curve of the regional electric vehicle total load provided by the embodiment of the present invention and the actual regional electric vehicle total load curve Figure 3 ;

[0035] Fig. 9 Schematic diagram of the comparison between the virtual collection curve of the regional electric vehicle total load provided by the embodiment of the present invention and the actual regional electric vehicle total load curve Figure 4 ;

[0036] Fig.10It is a schematic diagram of a regional electric vehicle charging load virtual collection device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0038] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0039] See also Figure 1 An embodiment of the present invention provides a method for virtual collection of regional electric vehicle charging loads, comprising:

[0040] S1. Linear interpolation is performed to fill in the missing values ​​of the historical data of electric vehicle load recorded by each electric meter in the selected area, and the historical load of electric vehicles in each area is obtained according to the information recorded by the electric meter to obtain the historical data set of electric vehicle load in each area and the historical data of the total electric vehicle load in the area;

[0041] S2. Use a dynamic time warping algorithm to capture the similarity of the time series of the historical data of the electric vehicle load in each area and the historical data of the total electric vehicle load in the region, and use it as the basis for screening the benchmark area; S3. Calculate the similarity of the time series between the selected benchmark areas to obtain a similarity matrix, use a hierarchical clustering method to cluster and reduce the dimension of the similarity matrix, smooth various types of data, and use the smoothed electric vehicle historical load data as the feature input of the neural network;

[0042] S4, building a CNN-GRU hybrid neural network model, using the features and the historical data of the regional total electric vehicle load as a training set, training the neural network model, and obtaining a regional electric vehicle charging load virtual collection model;

[0043] S5. Collect the electric vehicle meter data of the selected benchmark substation in real time, and use the clustering method and smoothing processing results as the input of the regional electric vehicle charging load virtual collection model to obtain the real-time value of the regional electric vehicle load and realize the virtual collection of the regional electric vehicle total load.

[0044] The structure diagram of the overall regional electric vehicle charging load virtual collection method is as follows: Figure 2 shown.

[0045] In one embodiment, the electric vehicle load historical data set of each substation and the regional total electric vehicle load historical data are obtained, including:

[0046] The power information recorded by the electric vehicle meters in each area is accumulated to obtain the historical data of the electric vehicle load in each area. The formula is:

[0047] ;

[0048] in, For Taiwan No. Electricity meter in The load power data of its electric vehicles at all times; For Taiwan The total number of electric meters belonging to electric vehicles; For Taiwan Next The total power data of its electric vehicle load at the moment.

[0049] According to the historical data of the electric vehicle load in each substation, the historical data of the total electric vehicle load in the region is obtained by cumulative summation, and the formula is as follows:

[0050] ;

[0051] in, The total electric vehicle load in the region Power data at the moment; The total number of downstage areas for the selected area.

[0052] In the embodiment of the present invention, a total of 692 substations and their electric vehicle cluster historical load data are selected in a residential area, and the historical data for one month is extracted at intervals of 15 minutes. The electric vehicle load historical record curves of each meter in each substation are accumulated to obtain a total of 692 substations. The electric vehicle charging load historical data of the 692 substations are summed to obtain the total electric vehicle load historical data in the residential area.

[0053] In one embodiment, the basis for screening benchmark areas includes:

[0054] The dynamic time warping algorithm is used to calculate the similarity between the historical data of electric vehicle load in each area and the historical data of total electric vehicle load in the region, and the shortest distance between the sequences is obtained. The area with the smallest sequence distance to the total electric vehicle load data in the region is selected as the benchmark area. The formula is as follows:

[0055] ;

[0056] in, For Taiwan Series data of electric vehicle load power; is the series data of regional total electric vehicle load power; Indicates that the dynamic time warping algorithm is used to calculate the similarity of two sequence data.

[0057] In an embodiment of the present invention, similarity calculation is performed on the historical data of electric vehicle load of the obtained 692 substations and the historical data of total electric vehicle load of the residential area, and the shortest distance between each substation sequence and the total load sequence is obtained. The substations are sorted in ascending order according to the minimum distance, and the substations with the highest sorting are selected as benchmark substations.

[0058] The aim is to reduce the computational burden of long-time scale processing of historical data. The present invention sets a specific time limit, such as 00:00 every day, to divide the continuous historical load data set into multiple independent data subsets. Subsequently, similarity analysis is performed independently on each data subset to calculate the similarity index within it. Finally, the similarity indexes of all subsets are accumulated and summed to obtain the sum of similarities of the overall data set.

[0059] In one embodiment, various types of data are smoothed, including:

[0060] According to the result of clustering and reducing the dimension of the similarity matrix using the hierarchical clustering method, the historical load data of the electric vehicle load in each substation in each clustered category is added together to reduce redundant information and reduce the complexity of subsequent calculations. The formula is as follows:

[0061] ;

[0062] in, is the total number of clusters of the hierarchical clustering method; The hierarchical clustering method is clustered kind; The clustering result is The total number of benchmark areas under the category; The clustering result is Class The series data of electric vehicle load power in each substation; The clustering result is The result of class smoothing on the sequence data.

[0063] In the embodiment of the present invention, hierarchical clustering is performed on the electric vehicle load data of the top 70 distribution areas, and the electric vehicle load historical data of each distribution area in each category are added together for smoothing.

[0064] In one embodiment, training a neural network model includes:

[0065] CNN-GRU (a combination of convolutional neural network and gated recurrent unit) is a hybrid neural network structure. The convolution layer of CNN is used to extract the characteristics and patterns of the smoothed historical load data of electric vehicles, and the pooling layer of CNN is used to reduce the output dimension of the convolution layer. Then, the GRU layer is used to learn the time series characteristics of the change of electric vehicle load power. The input data passes through the input layer, convolution layer, pooling layer, GRU layer, fully connected layer, and then outputs the regional level electric vehicle charging load virtual collection results through the output layer. The number of layers used here is single layer, and its structure is as follows: Figure 3 As shown. The internal formula of the GRU layer is as follows:

[0066] 1. Update gate: ;

[0067] in, is the input of the current GRU node; The hidden state information retained by the node last time; It is the weight matrix of the update gate, which is used to control the influence of the previous hidden state and the current input on the update gate; is the sigmoid function; To update the gate, decide how much past information should be retained at the current time step.

[0068] 2. Reset the gate: ;

[0069] in, It is the weight matrix of the reset gate, which is used to control the influence of the previous hidden state and the current input on the reset gate.

[0070] To reset the gate, decide how much past information should be ignored at the current time step.

[0071] 3. Candidate hidden states: ;

[0072] in, is the weight matrix of the candidate hidden states; is the hyperbolic tangent function; is the potential hidden state at the current time step, which will be adjusted by the output of the update gate.

[0073] 4. Final hidden state: ;

[0074] in, is the final hidden state at the current time step, which is the output of this GRU node.

[0075] In this embodiment, the specific parameters of the CNN-GRU hybrid neural network training adopted are: 16 filters are set in the convolution layer, the convolution kernel size is set to 8, the pooling layer window is set to 2, and the GRU layer is set to 48 neurons. To prevent the model from overfitting, the neuron output rejection index is set to 30%, 30% of the neuron output is randomly discarded, and finally the output is connected with the fully connected layer. The number of samples for each batch training is 16, and the sample order is not disrupted. The entire model is trained for 1000 rounds.

[0076] In one embodiment, the virtual collection of regional electric vehicle total load is implemented, including:

[0077] The real-time data of electric vehicle load in the selected benchmark substation is clustered according to the clustering results obtained from the training set, and is input into the electric vehicle load virtual acquisition model in real time after smoothing. The real-time output of the model is the virtual acquisition value of the total electric vehicle load at the regional level.

[0078] To demonstrate the superiority of this embodiment, consider the following four cases:

[0079] Case 1: Based on the dynamic time warping method, 10 substations are selected as benchmark substations to input the electric vehicle load virtual acquisition model.

[0080] Case 2: Based on the dynamic time warping method, 50 substations are selected as benchmark substations to input into the electric vehicle load virtual acquisition model.

[0081] Case 3: Based on the dynamic time warping method, 100 substations are selected as benchmark substations to input into the electric vehicle load virtual collection model.

[0082] Case 4: 100 substations were selected according to the dynamic time warping method and clustered into 10 categories using hierarchical clustering. The sum of each category was input into the electric vehicle load virtual acquisition model.

[0083] In Case 4, the results of selecting 100 substations by dynamic time warping and using hierarchical clustering are shown in the figure below. Figure 4 As shown, the horizontal axis represents different samples, and the vertical axis represents the cluster distance. Here, the number of clusters is selected as 10.

[0084] Table 1 Comparison of evaluations of different models

[0085]

[0086] It can be seen from Table 1 that when the number of selected benchmark areas gradually increases, the error between the virtual collection value and the real value of the electric vehicle load virtual collection model becomes smaller. However, when the number of input benchmark areas is too large, the dimensionality disaster makes it difficult for the model to capture effective patterns and correlations. The clustering dimension reduction and smoothing processing method described in the present invention improves the model performance. The virtual collection effects of the four cases proposed in the embodiments of the present invention are as follows: Figure 6 Figure 7 , Figure 8 and Fig. 9 The evaluation comparison results of different schemes are shown in Figure 5 As shown, it is only necessary to select a small number of substations in the region as benchmark substations and collect the electric vehicle loads of the benchmark substations in real time, so as to accurately collect the electric vehicle load data of the entire region virtually, which proves the effectiveness of the present invention.

[0087] In summary, when it is impossible to obtain a wide range of massive electric vehicle load data in real time, the present invention first obtains the electric vehicle load historical data set of each substation and the historical data of the total electric vehicle load in the region. Then, the dynamic time warping algorithm is used to screen the benchmark substation, and the hierarchical clustering and smoothing methods are used to reduce the complexity of the model. Then, the constructed CNN-GRU neural network model is trained using historical data to obtain a regional electric vehicle charging load virtual acquisition model. Finally, the electric vehicle load data of the benchmark substation is collected in real time to complete the virtual acquisition of the electric vehicle load in the entire region. In this embodiment, the regional electric vehicle charging load virtual acquisition method proposed by the present invention shows superior performance, which verifies the feasibility and effectiveness of the present invention.

[0088] Corresponding to the aforementioned embodiment of a regional electric vehicle charging load virtual collection method, the present invention also provides an embodiment of a regional electric vehicle charging load virtual collection system. The system includes a load collection module, a benchmark station area screening module, a data processing module and a virtual collection module;

[0089] The load acquisition module is used to obtain the historical data set of electric vehicle load in each substation and the historical data of the total electric vehicle load in the region, and subsequently collect the electric vehicle meter data of the selected benchmark substation in real time;

[0090] The benchmark area screening module uses a dynamic time warping algorithm to capture the similarity of the time series of the historical data of the electric vehicle load in each area and the historical data of the total electric vehicle load in the region; and uses the time series similarity to screen the benchmark area;

[0091] The data processing module is used to calculate the similarity matrix of the time series similarity of each benchmark station area, cluster and reduce the dimension of the similarity matrix using a hierarchical clustering method, and perform smoothing on various types of data.

[0092] The virtual acquisition module is used to build a CNN-GRU hybrid neural network model. The features and historical data of the total electric vehicle load in the region are used as training sets to train the neural network model to obtain a regional electric vehicle charging load virtual acquisition model. Clustering and smoothing processing are performed based on the real-time collected electric vehicle meter data of the selected benchmark substation. The results are used as the input of the regional electric vehicle charging load virtual acquisition model to obtain the real-time value of the regional electric vehicle load.

[0093] Corresponding to the aforementioned embodiment of a method for virtual collection of charging load of a regional electric vehicle, the present invention further provides an embodiment of a device for virtual collection of charging load of a regional electric vehicle.

[0094] See also Fig.10 An embodiment of the present invention provides a regional electric vehicle charging load virtual collection device, including a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, it is used to implement a regional electric vehicle charging load virtual collection method in the above embodiment.

[0095] The embodiment of the regional electric vehicle charging load virtual acquisition device provided by the present invention can be applied to any device with data processing capabilities, and the device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the internal memory for execution. From the hardware level, if Fig.10 As shown, it is a hardware structure diagram of a device with data processing capability for a regional electric vehicle charging load virtual collection device provided by the present invention, except Fig.10 In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in which the apparatus in the embodiments is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.

[0096] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0097] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0098] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a method for virtual collection of regional electric vehicle charging loads in the above embodiment is implemented.

[0099] The computer-readable storage medium may be an internal storage unit of any device with data processing capability described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium may also be an external storage device of any device with data processing capability, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capability. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capability, and may also be used to temporarily store data that has been output or is to be output.

[0100] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0101] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0102] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0103] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0104] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. The acquisition, storage, use, and processing of data involved in the embodiments of this application are in compliance with the relevant provisions of national laws and regulations.

[0105] The above are preferred embodiments of the present invention. It should be noted that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A regional electric vehicle load virtual collection method, characterized in that: The following steps are involved: S1. Obtain the historical data set of electric vehicle load in each substation and the historical data of the total electric vehicle load in the region; S2. Use the dynamic time warping algorithm to capture the similarity between the historical data of electric vehicle load in each area and the time series of the historical data of the regional total electric vehicle load. Specifically, use the dynamic time warping algorithm to calculate the similarity, obtain the shortest distance between the sequences, sort the shortest distances, and select several areas with the shortest distance from the regional total electric vehicle load data sequence as benchmark areas. The formula is as follows: d i =dtw(P i ,P tol ); Among them, P i is the sequence data of electric vehicle load power in area i; P tol is the sequence data of the total electric vehicle load power in the region; dtw(,) indicates that the dynamic time warping algorithm is used to calculate the similarity of the two sequence data; Use time series similarity to select benchmark areas; S3, calculating the similarity matrix of the time series similarity of each benchmark area, clustering and reducing the dimension of the similarity matrix by using a hierarchical clustering method, and performing smoothing processing on various types of data to obtain the input of the neural network model; the smoothing processing is specifically: according to the result of clustering and reducing the dimension of the similarity matrix, adding the historical load data of the electric vehicle load under each area in each category obtained by clustering; S4, building a CNN-GRU hybrid neural network model, taking the features and the historical data of the regional total electric vehicle load as the training set, training the neural network model, and obtaining a regional electric vehicle load virtual collection model; S5. Collect electric vehicle meter data of the selected benchmark substation in real time, and use the hierarchical clustering method and smoothing processing results as input of the regional electric vehicle load virtual collection model to obtain the real-time value of the regional electric vehicle load.

2. A method for virtual collection of regional electric vehicle loads according to claim 1, characterized in that: The S1 specifically includes: linearly interpolating and filling the missing values ​​of the historical data of electric vehicle load recorded by each meter in the selected area, accumulating the power information recorded by the electric vehicle meters in each substation to obtain the historical data of the electric vehicle load in each substation, and accumulating and summing the historical data of the electric vehicle load in each substation to obtain the historical data of the total electric vehicle load in the region.

3. A method for virtual collection of regional electric vehicle loads according to claim 1, characterized in that: In the similarity calculation, the continuous historical load data set is divided into multiple independent data subsets, and similarity analysis is performed independently on each data subset to calculate the similarity index within it; Finally, the similarity indices of all subsets are accumulated and summed to obtain the total similarity of the entire data set.

4. A method for virtual collection of regional electric vehicle loads according to claim 1, characterized in that: The CNN-GRU hybrid neural network model specifically includes: using the convolution layer of CNN to extract the characteristics and patterns of the smoothed electric vehicle historical load data, and using the pooling layer of CNN to achieve output dimensionality reduction of the convolution layer, and then using the GRU layer to learn the time series characteristics of electric vehicle load power changes. Its input data passes through the input layer, convolution layer, pooling layer, GRU layer, and fully connected layer, and then outputs the regional electric vehicle load virtual acquisition results through the output layer.

5. A method for virtual collection of regional electric vehicle loads according to claim 4, characterized in that: A neuron output discarding index is set in the GRU layer. When the GRU layer performs output, output discarding is performed according to the neuron output discarding index to prevent overfitting.

6. A regional electric vehicle load virtual collection system for implementing the method according to any one of claims 1 to 5, characterized in that: The system includes a load acquisition module, a benchmark area screening module, a data processing module and a virtual acquisition module; The load acquisition module is used to obtain the historical data set of electric vehicle load in each substation and the historical data of the total electric vehicle load in the region, and subsequently collect the electric vehicle meter data of the selected benchmark substation in real time; The benchmark area screening module uses a dynamic time warping algorithm to capture the similarity of the time series of the historical data of the electric vehicle load in each area and the historical data of the total electric vehicle load in the region; and uses the time series similarity to screen the benchmark area; The data processing module is used to calculate the similarity matrix of the time series similarity of each benchmark station area, cluster and reduce the dimension of the similarity matrix using a hierarchical clustering method, and perform smoothing on various types of data. The virtual acquisition module is used to build a CNN-GRU hybrid neural network model, and uses the features and historical data of the total regional electric vehicle load as training sets to train the neural network model to obtain a regional electric vehicle load virtual acquisition model. Clustering and smoothing processing are performed based on the real-time collected electric vehicle meter data of the selected benchmark substation, and the results are used as the input of the regional electric vehicle load virtual acquisition model to obtain the real-time value of the regional electric vehicle load.

7. A regional electric vehicle load virtual collection device, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, a regional electric vehicle load virtual collection method according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, a regional electric vehicle load virtual collection method according to any one of claims 1 to 5 is implemented.

Citation Information

Patent Citations

  • Electric vehicle charging load prediction method and system based on FCM clustering and LSTM

    CN113988426A

  • Regional electric vehicle charging load prediction method

    CN115907122A

  • Electric vehicle charging supply control method and device, terminal and medium

    CN116872780A

  • Prediction method for charging loads of electric vehicles with consideration of data correlation

    US20230074700A1