Charging load prediction method for electric vehicle charging facility

Abstract

The invention relates to the technical field of charging station prediction, and discloses an electric vehicle charging facility charging load prediction method, which comprises the following steps: calculating the capacity of a charging facility; historical charging features are obtained and input into the improved BiLSTM model to obtain a first charging facility load prediction value; inputting the traffic flow real-time information near the charging station as input data into a preset XGBoost model to obtain a second charging facility load prediction value; calculating a total charging load prediction value of the electric vehicle charging facilities according to the first charging facility load prediction value and the second charging facility load prediction value; and performing start-stop adjustment on the charging facilities of the charging station according to the charging facility capacity and the total charging load predicted value. According to the method, the historical charging characteristics and the traffic flow real-time information near the charging station are input into the improved BiLSTM model and the XGBoost model to carry out charging facility load prediction, the charging load of the charging station can be accurately predicted, and the accurate prediction of the charging load of the charging station is realized on the premise of fully considering the real-time change of the input data.

Description

Technical Field

[0001] The present invention relates to the technical field of charging station prediction, and in particular to a method for predicting charging load of an electric vehicle charging facility. Background Art

[0002] With the rapid development of the new energy vehicle industry, the number of electric vehicles continues to grow, and the demand for charging facilities has also increased. In the context of promoting green travel and sustainable development, the rational layout and scientific planning of charging infrastructure are the key to the large-scale promotion of electric vehicles. However, when forecasting the load of charging facilities, different levels and different scenarios are not considered, resulting in inaccurate load forecast values ​​and unreasonable resource allocation. The current load characteristic extraction methods include the index mean selection method, the symbol segmentation aggregation method, the daily maximum load selection method, and the non-parametric kernel density estimation method.

[0003] In general, there are still many shortcomings in the existing power planning methods in terms of charging facility classification and charging load prediction of charging facilities. Therefore, a charging load prediction method for electric vehicle charging facilities is proposed to solve the above problems. Summary of the invention

[0004] The present invention provides a method for predicting the charging load of electric vehicle charging facilities. The method inputs historical charging characteristics and real-time information of traffic flow near charging stations into improved BiLSTM models and XGBoost models to predict the charging facility load. The method can accurately predict the charging load of charging stations, and realize accurate prediction of the charging load of charging stations under the premise of fully considering the real-time changes of input data.

[0005] The present invention provides a method for predicting charging load of an electric vehicle charging facility, comprising:

[0006] Analyze the charging facilities of electric vehicle charging stations and calculate the capacity of charging facilities;

[0007] Obtaining historical charging characteristics, and inputting the historical charging characteristics into an improved BiLSTM model to obtain a load forecast value of a first charging facility;

[0008] Using real-time traffic flow information near the charging station as input data, and inputting the input data into a preset XGBoost model to obtain a second charging facility load prediction value;

[0009] Calculate the total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value;

[0010] The charging facilities of the charging station are started and stopped according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

[0011] Furthermore, the step of obtaining historical charging characteristics and inputting the historical charging characteristics into the improved BiLSTM model to obtain the load forecast value of the first charging facility includes:

[0012] The historical charging load, weather, temperature, date and other data of the charging station are used as input to construct historical charging load statistics and new features related to the date as historical charging features;

[0013] The SSA optimization algorithm is used to continue processing the historical charging load data, and the penalty factor α and the modal number K of the VMD model are determined after iteration;

[0014] Perform VMD decomposition on the historical charging load data to obtain K modal components, among which the non-periodic modal component IMF1 enters the middle layer to obtain the influencing characteristic factors, and then enters the BiLSTM layer together with the periodic modal components IMF2 and IMFK;

[0015] In the BiLSTM layer, the non-periodic modal components are predicted based on characteristic factors, and the periodic modal components are predicted directly based on historical charging load data;

[0016] Entering the output layer, all prediction results are modally reconstructed to obtain the load prediction value of the first charging facility.

[0017] Furthermore, the step of continuing to process the historical charging load data using the SSA optimization algorithm and determining the penalty factor α and the modal number K of the VMD model after iteration includes:

[0018] VMD determines the number of modes, and then adaptively matches the center frequency and limited bandwidth according to the number of signal decompositions. Through Gaussian demodulation, the signal is decomposed into a series of modal component signals with multiple different frequencies, strong time sequence and relatively stable amplitude, and the noise of the signal is concentrated in the first mode.

[0019] SSA is used to solve the optimal values ​​of the penalty factor α and the mode number K in the two VMD models.

[0020] Furthermore, the step of using the real-time traffic flow information near the charging station as input data and inputting the input data into a preset XGBoost model to obtain a second charging facility load prediction value includes:

[0021] Traffic flow data is calculated by Baidu Maps based on massive traffic travel big data, vehicle trajectory big data, and location positioning big data, including congestion index and average speed;

[0022] The traffic flow data is input into the preset XGBoost model to obtain the load forecast value of the second charging facility; wherein, XGBoost aims to reduce the residual, continuously constructs sub-models and generates forecast values ​​until the residual can meet the error requirements.

[0023] Furthermore, in the traffic flow data, the calculation formula of the congestion index is:

[0024]

[0025] Where η is the congestion index; t s is the actual travel time of the vehicle; t c is the vehicle smooth travel time; c v is the average speed when the traffic is unimpeded; s is the actual average speed of the vehicle.

[0026] Furthermore, the step of calculating the total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value comprises:

[0027] Subtract the second charging facility load prediction value from the second charging facility load prediction value to obtain a charging facility prediction residual value;

[0028] The charging facility prediction residual value is added to the first charging facility load prediction value to obtain the total charging load prediction value of the electric vehicle charging facility.

[0029] Furthermore, the step of starting and stopping the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities includes:

[0030] When the capacity of the charging facility is greater than the predicted value of the total charging load of the electric vehicle charging facility, all the charging facilities of the charging station are turned on for standby charging;

[0031] When the capacity of the charging facility is less than or equal to the predicted value of the total charging load of the electric vehicle charging facility, the use of some charging facilities is stopped according to the capacity of a single charging facility.

[0032] The present invention also provides a charging load prediction device for an electric vehicle charging facility, comprising:

[0033] The first calculation module is used to analyze the charging facilities of the electric vehicle charging station and calculate the capacity of the charging facilities;

[0034] A first prediction module, used to obtain historical charging characteristics, and input the historical charging characteristics into an improved BiLSTM model to obtain a first charging facility load prediction value;

[0035] A second prediction module is used to take the real-time information of traffic flow near the charging station as input data, and input the input data into a preset XGBoost model to obtain a second charging facility load prediction value;

[0036] A second calculation module, used to calculate a total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value;

[0037] The adjustment module is used to start and stop the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

[0038] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0039] The present invention also provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the steps of the above method when executed by a processor.

[0040] The beneficial effects of the present invention are:

[0041] The present invention inputs historical charging characteristics and real-time information of traffic flow near charging stations into different models for prediction, fully utilizing the real-time changes of data, and adopts improved BiLSTM model and XGBoost model to predict the charging facility load of different data. The improved BiLSTM model performs model training on features with small real-time changes such as weather and temperature, and the XGBoost model performs model improvement on traffic flow data with large real-time changes, so as to better adapt to the real-time changes of environmental factors. The charging load prediction range is refined to the unit of charging station, so as to accurately predict the charging load of the charging station, and realize accurate prediction of the charging load of the charging station under the premise of fully considering the real-time changes of input data. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The figure is a schematic diagram of a method flow according to an embodiment of the present invention.

[0043] Figure 2 FIG. 1 is a schematic diagram of a device structure according to an embodiment of the present invention.

[0044] Figure 3 The figure is a schematic diagram of the internal structure of a computer device according to an embodiment of the present invention.

[0045] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0046] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0047] like Figure 1 As shown, the present invention provides a method for predicting charging load of electric vehicle charging facilities, comprising:

[0048] S1. Analyze the charging facilities of electric vehicle charging stations and calculate the capacity of charging facilities.

[0049] Combined with relevant planning data, the regional electric vehicle charging facilities and future charging facilities are analyzed. The charging facilities are then divided into levels and divided into different scenarios to facilitate subsequent prediction and analysis of the charging load of the charging facilities.

[0050] S2. Obtain historical charging characteristics, and input the historical charging characteristics into an improved BiLSTM model to obtain a load forecast value of the first charging facility.

[0051] Step S2 specifically includes:

[0052] S21, taking the historical charging load, weather, temperature, date and other data of the charging station as input, constructing historical charging load statistics and new features related to the date as historical charging features;

[0053] S22. Use the SSA optimization algorithm to continue processing the historical charging load data, and determine the penalty factor α and the modal number K of the VMD model after iteration.

[0054] VMD is an adaptive, completely non-recursive modal variation and signal processing method that can determine the number of modes, and then adaptively match the center frequency and limited bandwidth according to the number of signal decompositions. Through Gaussian demodulation, the signal is decomposed into a series of modal component signals with multiple different frequencies, strong timing and relatively stable amplitudes, and the noise of the signal is concentrated in the first mode. VMD is divided into two parts: construction and solution of variational problems.

[0055] The construction of the variational problem includes: using the original data of the power load as input data, and the decomposed modal components as output data, demodulating the signal through Gaussian smoothing to obtain the bandwidth description of each mode. The signal is demodulated through the square root of the norm gradient (i.e., Gaussian smoothing filtering) to obtain the bandwidth description of each mode. The role of Gaussian smoothing filtering is to make the signal more stable and reduce the impact of noise. After VMD decomposition, the noise of the signal is concentrated on the first modal component, and finally a non-periodic main modal component and multiple smooth periodic modal components are generated.

[0056] The solution of variational problem includes: The solution of variational problem is regarded as the conditional minimum of multivariate function. a) Using quadratic penalty factor α and Lagrange multiplier λ, the constrained variational problem is transformed into an unconstrained variational problem. b) Using alternating multiplier method to search for saddle point and continuously update constraint parameters. c) Repeatedly update the parameter value until the iteration constraint is satisfied.

[0057] The two important parameters that affect the accuracy of VMD model results are the penalty factor α and the modal number K. SSA is used to solve the optimal values ​​of the penalty factor α and the modal number K in these two VMD models.

[0058] S23, perform VMD decomposition on the historical charging load data to obtain K modal components, among which the non-periodic modal component IMF1 enters the middle layer to obtain the influencing characteristic factors, and then enters the BiLSTM layer together with the periodic modal components IMF2 and IMFK;

[0059] S24, in the BiLSTM layer, predicting the non-periodic modal component based on the characteristic factors, and predicting the periodic modal component directly based on the historical charging load data;

[0060] BiLSTM is composed of forward LSTM and backward LSTM. LSTM is a special recurrent neural network. When processing long time series, the LSTM method can effectively retain the forward and backward dependencies of historical data and solve the problem of gradient disappearance to a certain extent. BiLSTM adopts a more complex bidirectional structure, which can better capture the short-term bidirectional implicit dependency of data. The BiLSTM model is used to predict the load of the modal components of the power load data after VMD decomposition. Since the prediction is made separately, the prediction principle of a single mode is analyzed, and the original load data is. The model structure of LSTM is mainly composed of three gate control units that control information transmission: forget gate, input gate, and output gate.

[0061] The first part is the forget gate, which is used to process the unimportant information in the previous information. Among them, e(t) is the output vector of the forget gate, h(t-1) is the output of the previous LSTM network node, and x(t) is the input of the current network node, including the power load data u(t) and feature data at time t (only the first mode needs to introduce feature data); We is the forget layer parameter matrix, which represents the matrix weight; σ is the Sigmoid function, which compresses the input signal so that each element is in the interval [0,1]; be is the forget gate bias. The output result of the forget gate is e(t) = σ(We · [h(t-1), x(t)] + be).

[0062] The second part is the input gate, which is used to store the information of this node. g(t) is all the information that can be stored, and it is also compressed by the signal of the σ function. C(t) is the candidate vector, which is combined with g(t) to create updated information. Tanh is the hyperbolic tangent function, which compresses the vector elements to the interval [-1,1]. Wg and Wc are the input layer matrix weights. bg and bc are the input layer biases. The output of the input gate is: g(t) = σ(Wg · [h(t-1), x(t)] + bg), C(t) = tanh(Wc · [h(t-1), x(t)] + bc).

[0063] After completing the output of the forget gate and the input gate, the current state value can be updated. C(t-1) is the previous state value, and C(t) is the current state value. After receiving the signals from the forget gate and the input gate, the update result is C1(t)=e(t)·C(t-1)+g(t)·C(t).

[0064] The last part is the output gate. Wo and bo are the weight matrix and bias signal of the output gate respectively. The output result of the output gate is: o(t)=σ(Wo·[h(t-1),x(t)]+bo), h(t)=o(t)·tanh(C(t)).

[0065] The forward LSTM and the backward LSTM are combined into BiLSTM. According to the above conclusion, assuming that the original sequence of the forward LSTM is {S(0), S(1), …, S(m)}, and the output vector generated by the LSTM network is {h(0), h(1), …, h(m)}, then the corresponding reverse sequence of the backward LSTM is {S(m), S(m-1), …, S(0)}, and the m vectors generated by the LSTM network are {h'(m), h'(m-1), …, h'(0)}. Therefore, the prediction result of the BiLSTM model at time t is: h(t) = o(t) tanh(C(t)) + o'(mt) tanh(C'(mt)). In the formula, o'(mt) and C'(mt) are the temporary variables and state values ​​generated by the reverse sequence through the LSTM network, respectively.

[0066] S25, enter the output layer, reconstruct all prediction results modally, and obtain the first charging facility load prediction value. The above is the prediction result of a single mode. After predicting the data of each mode, data reconstruction is required. The relationship between the total prediction ha(t) of the charging station and the prediction result hi(t) of each mode is obtained as follows:

[0067] S3. Using the real-time traffic flow information near the charging station as input data, and inputting the input data into a preset XGBoost model to obtain a second charging facility load prediction value;

[0068] Step S3 specifically includes:

[0069] S31. Traffic flow data is calculated by Baidu Maps based on massive traffic travel big data, vehicle trajectory big data, and location positioning big data, including congestion index and average speed. The calculation formula for the congestion index is:

[0070]

[0071] Where η is the congestion index; t s is the actual travel time of the vehicle; t c is the vehicle smooth travel time; c v is the average speed when the traffic is unimpeded; s is the actual average speed of the vehicle.

[0072] S32. Input the traffic flow data into the preset XGBoost model to obtain the load prediction value of the second charging facility; wherein, XGBoost aims to reduce the residual, continuously constructs sub-models and generates prediction values ​​until the residual can meet the error requirement.

[0073] The prediction accuracy of the XGBoost model increases with the number of iterations, and it is widely used in classification tasks and regression problems. Compared with GBDT and random forest, the XGBoost model has made many improvements. By adding regularization terms, the complexity of the decision tree is effectively controlled, the generalization ability of the model is improved, and the risk of overfitting is reduced. At the same time, the XGBoost model can realize random sampling and automatic processing of missing values, support second-order Taylor expansion to remove constant terms in the objective function, select parallel features, and customize loss functions, effectively improving the prediction speed.

[0074] S4. Calculate the total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value;

[0075] Step S4 includes:

[0076] S41, subtracting the second charging facility load prediction value from the second charging facility load prediction value to obtain a charging facility prediction residual value;

[0077] S42: Add the charging facility prediction residual value to the first charging facility load prediction value to obtain a total charging load prediction value of the electric vehicle charging facility.

[0078] S5. Start and stop the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

[0079] Step S5 specifically includes:

[0080] S51, when the capacity of the charging facility is greater than the predicted value of the total charging load of the electric vehicle charging facility, all the charging facilities of the charging station are turned on for standby charging;

[0081] S52: When the capacity of the charging facility is less than or equal to the predicted value of the total charging load of the electric vehicle charging facility, stop using some charging facilities according to the capacity of a single charging facility.

[0082] The present invention inputs historical charging characteristics and real-time information of traffic flow near charging stations into different models for prediction, fully utilizing the real-time changes of data, and adopts improved BiLSTM model and XGBoost model to predict the charging facility load of different data. The improved BiLSTM model performs model training on features with small real-time changes such as weather and temperature, and the XGBoost model performs model improvement on traffic flow data with large real-time changes, so as to better adapt to the real-time changes of environmental factors. The charging load prediction range is refined to the unit of charging station, so as to accurately predict the charging load of the charging station, and realize accurate prediction of the charging load of the charging station under the premise of fully considering the real-time changes of input data.

[0083] like Figure 2 As shown, the present invention also provides a charging load prediction device for an electric vehicle charging facility, comprising:

[0084] The first calculation module 1 is used to analyze the charging facilities of the electric vehicle charging station and calculate the capacity of the charging facilities;

[0085] The first prediction module 2 is used to obtain historical charging characteristics and input the historical charging characteristics into the improved BiLSTM model to obtain a first charging facility load prediction value;

[0086] A second prediction module 3 is used to take the real-time information of traffic flow near the charging station as input data, and input the input data into a preset XGBoost model to obtain a second charging facility load prediction value;

[0087] A second calculation module 4, used to calculate a total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value;

[0088] The adjustment module 5 is used to start and stop the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

[0089] In one embodiment, the first prediction module 2 includes:

[0090] A construction unit, used to take the historical charging load, weather, temperature, date and other data of the charging station as input, and construct historical charging load statistics and new features related to the date as historical charging features;

[0091] A processing unit, used to continue processing the historical charging load data using an SSA optimization algorithm, and to determine the penalty factor α and the modal number K of the VMD model after iteration;

[0092] The decomposition unit is used to perform VMD decomposition on the historical charging load data to obtain K modal components, among which the non-periodic modal component IMF1 enters the middle layer to obtain the influencing characteristic factors, and then enters the BiLSTM layer together with the periodic modal components IMF2 and IMFK;

[0093] The first prediction unit is used to predict the non-periodic modal component based on the characteristic factors in the BiLSTM layer, and predict the periodic modal component directly based on the historical charging load data;

[0094] The output unit is used to enter the output layer, reconstruct all prediction results modally, and obtain the first charging facility load prediction value.

[0095] In one embodiment, the processing unit includes:

[0096] Determine the subunit, which is used for VMD to determine the number of modes, and then adaptively match the center frequency and limited bandwidth according to the number of signal decompositions. Decompose the signal into a series of modal component signals with multiple different frequencies, strong timing and relatively stable amplitudes through Gaussian demodulation, and concentrate the signal noise in the first mode.

[0097] The solving subunit is used to solve the optimal values ​​of the penalty factor α and the mode number K in the two VMD models using SSA.

[0098] In one embodiment, the second prediction module 3 includes:

[0099] Acquisition unit, traffic flow data is calculated and mined by Baidu Maps based on massive traffic travel big data, vehicle trajectory big data and location positioning big data, including congestion index and average speed;

[0100] The second prediction unit is used to input the traffic flow data into a preset XGBoost model to obtain a second charging facility load prediction value; wherein XGBoost aims to reduce the residual, continuously constructs sub-models, and generates prediction values ​​until the residual can meet the error requirement.

[0101] In one embodiment, in the acquisition unit, the calculation formula of the congestion index is:

[0102]

[0103] Where η is the congestion index; t s is the actual travel time of the vehicle; t c is the vehicle smooth travel time; c v is the average speed when the traffic is unimpeded; s is the actual average speed of the vehicle.

[0104] In one embodiment, the second calculation module 4 includes:

[0105] a subtraction unit, configured to subtract the second charging facility load prediction value from the second charging facility load prediction value to obtain a charging facility prediction residual value;

[0106] The adding unit is used to add the charging facility prediction residual value to the first charging facility load prediction value to obtain the total charging load prediction value of the electric vehicle charging facility.

[0107] In one embodiment, the adjustment module 5 includes:

[0108] An opening unit, used to open all charging facilities of the charging station for standby charging when the capacity of the charging facility is greater than the predicted value of the total charging load of the electric vehicle charging facility;

[0109] The stopping unit is used to stop the use of some charging facilities according to the capacity of a single charging facility when the capacity of the charging facility is less than or equal to the predicted value of the total charging load of the electric vehicle charging facility.

[0110] The above modules, units and sub-units are used to execute the corresponding steps in the above electric vehicle charging facility charging load prediction method. The specific implementation method thereof is described in the above method embodiment and will not be repeated here.

[0111] like Figure 3 As shown, the present invention also provides a computer device, which can be a server, and its internal structure can be as shown in Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store all data required for the process of the electric vehicle charging facility charging load prediction method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the electric vehicle charging facility charging load prediction method is implemented.

[0112] Those skilled in the art will understand that Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0113] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, any one of the above-mentioned methods for predicting charging load of electric vehicle charging facilities is implemented.

[0114] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0115] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0116] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting charging load of electric vehicle charging facilities, characterized in that: include: Analyze the charging facilities of electric vehicle charging stations and calculate the capacity of charging facilities; Obtaining historical charging characteristics, and inputting the historical charging characteristics into an improved BiLSTM model to obtain a load forecast value of a first charging facility; Using real-time traffic flow information near the charging station as input data, and inputting the input data into a preset XGBoost model to obtain a second charging facility load prediction value; Calculate the total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value; The charging facilities of the charging station are started and stopped according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

2. The method for predicting charging load of electric vehicle charging facilities according to claim 1, characterized in that: The step of obtaining historical charging characteristics and inputting the historical charging characteristics into the improved BiLSTM model to obtain the load forecast value of the first charging facility includes: The historical charging load, weather, temperature, date and other data of the charging station are used as input to construct historical charging load statistics and new features related to the date as historical charging features; The SSA optimization algorithm is used to continue processing the historical charging load data, and the penalty factor α and the modal number K of the VMD model are determined after iteration; Perform VMD decomposition on the historical charging load data to obtain K modal components, among which the non-periodic modal component IMF1 enters the middle layer to obtain the influencing characteristic factors, and then enters the BiLSTM layer together with the periodic modal components IMF2 and IMFK; In the BiLSTM layer, the non-periodic modal components are predicted based on characteristic factors, and the periodic modal components are predicted directly based on historical charging load data; Entering the output layer, all prediction results are modally reconstructed to obtain the load prediction value of the first charging facility.

3. The method for predicting charging load of electric vehicle charging facilities according to claim 2, characterized in that: The step of continuing to process the historical charging load data using the SSA optimization algorithm and determining the penalty factor α and the modal number K of the VMD model after iteration includes: VMD determines the number of modes, and then adaptively matches the center frequency and limited bandwidth according to the number of signal decompositions. Through Gaussian demodulation, the signal is decomposed into a series of modal component signals with multiple different frequencies, strong time sequence and relatively stable amplitude, and the noise of the signal is concentrated in the first mode. SSA is used to solve the optimal values ​​of the penalty factor α and the mode number K in the two VMD models.

4. The method for predicting charging load of electric vehicle charging facilities according to claim 1, characterized in that: The step of using the real-time traffic flow information near the charging station as input data and inputting the input data into a preset XGBoost model to obtain a second charging facility load prediction value includes: Traffic flow data is calculated by Baidu Maps based on massive traffic travel big data, vehicle trajectory big data, and location positioning big data, including congestion index and average speed; The traffic flow data is input into the preset XGBoost model to obtain the load forecast value of the second charging facility; wherein, XGBoost aims to reduce the residual, continuously constructs sub-models and generates forecast values ​​until the residual can meet the error requirements.

5. The method for predicting charging load of electric vehicle charging facilities according to claim 1, characterized in that: In the traffic flow data, the calculation formula of the congestion index is: Where η is the congestion index; t s is the actual travel time of the vehicle; t c is the vehicle smooth travel time; c is the average speed when the traffic is unimpeded; v s is the actual average speed of the vehicle.

6. The method for predicting charging load of electric vehicle charging facilities according to claim 1, characterized in that: The step of calculating the total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value comprises: Subtract the second charging facility load prediction value from the second charging facility load prediction value to obtain a charging facility prediction residual value; The charging facility prediction residual value is added to the first charging facility load prediction value to obtain the total charging load prediction value of the electric vehicle charging facility.

7. The method for predicting charging load of electric vehicle charging facilities according to claim 1, characterized in that: The step of starting and stopping the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities includes: When the capacity of the charging facility is greater than the predicted value of the total charging load of the electric vehicle charging facility, all the charging facilities of the charging station are turned on for standby charging; When the capacity of the charging facility is less than or equal to the predicted value of the total charging load of the electric vehicle charging facility, the use of some charging facilities is stopped according to the capacity of a single charging facility.

8. A charging load prediction device for electric vehicle charging facilities, characterized in that: include: The first calculation module is used to analyze the charging facilities of the electric vehicle charging station and calculate the capacity of the charging facilities; A first prediction module, used to obtain historical charging characteristics, and input the historical charging characteristics into an improved BiLSTM model to obtain a first charging facility load prediction value; A second prediction module is used to take the real-time information of traffic flow near the charging station as input data, and input the input data into a preset XGBoost model to obtain a second charging facility load prediction value; A second calculation module, used to calculate a total charging load prediction value of the electric vehicle charging facility according to the first charging facility load prediction value and the second charging facility load prediction value; The adjustment module is used to start and stop the charging facilities of the charging station according to the capacity of the charging facilities and the predicted value of the total charging load of the electric vehicle charging facilities.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images