Electric vehicle charging load prediction system and method based on traffic flow

Abstract

The present invention discloses a system and method for predicting the charging load of an electric vehicle based on traffic flow, and relates to the field of orderly charging and discharging of electric vehicles. The system comprises a cloud monitoring and management platform, a road network data management subsystem, a traffic flow prediction and management subsystem, an on-board data acquisition and control subsystem, and a user management subsystem; wherein the cloud monitoring and management platform receives and displays relevant data and operating status of the road network data management subsystem, the traffic flow prediction and management subsystem, the on-board data acquisition and control subsystem, and the user management subsystem in real time, processes the data to obtain spatiotemporal load prediction information for charging of the electric vehicle, and transmits the information to a load prediction display interface. The present invention conducts comprehensive data mining on various factors that can affect the charging load of the electric vehicle, quantitatively analyzes the influence of various factors on the charging load of the electric vehicle, and makes the prediction of the charging load prediction model of the electric vehicle more accurate.

Description

Technical Field

[0001] The present invention belongs to the field of orderly charging and discharging of electric vehicles, and in particular relates to a system and method for predicting charging load of electric vehicles based on traffic flow. Background Art

[0002] With the energy crisis and global warming, electric vehicles (EVs) are gaining significant attention worldwide to reduce fossil fuel use and reduce vehicle emissions. EVs not only conserve fossil fuels but also produce no exhaust like gasoline-powered vehicles, achieving "zero emissions" and reducing environmental pollution. Therefore, the development of EVs is a future trend in the automotive industry, and countries around the world have successively issued policies to promote their development. With the development of the energy industry and the increasing severity of environmental pollution, EVs, as a new mode of transportation, offer significant advantages in reducing CO2 emissions and alleviating the energy crisis, attracting the attention of governments and scholars worldwide. According to the "Energy-Saving and New Energy Vehicle Industry Development Plan" issued by the State Council, the cumulative production and sales of pure electric vehicles and plug-in hybrid vehicles will exceed 5 million by 2020. However, the widespread adoption of EVs will significantly impact power system operations, exacerbating peak and valley load variations and increasing the difficulty of optimizing grid operation and control. As participants in transportation, EVs have significant impacts not only on transportation networks but also on power grids. Therefore, predicting the spatiotemporal distribution of electric vehicle charging load based on traffic flow forecasts and quantitatively analyzing the impact of electric vehicles on the power grid will provide important references for studying electric vehicle control strategy models and promoting the safe and stable operation of power systems. It will also lay a solid foundation for the site selection and layout planning of charging stations after large-scale electric vehicles enter the grid. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to solve the problem of excessive load on the distribution network caused by inaccurate load forecasting, and also to solve the problem of traffic congestion caused by large-scale electric vehicle charging.

[0004] The technical solution adopted in the present invention is as follows:

[0005] A traffic flow-based electric vehicle charging load prediction system includes a cloud monitoring management platform, a road network data management subsystem, a traffic flow prediction management subsystem, an on-board data acquisition and control subsystem, and a user management subsystem.

[0006] The vehicle-mounted data acquisition and control subsystem is used to collect driving data and road condition data of the electric vehicle participating in real-time traffic, process the driving data and road condition data to obtain first driving data and first road condition data, and upload the first driving data and the first road condition data to the road network data management subsystem;

[0007] The road network data management subsystem is configured to receive the first driving data and the first road condition data sent by the vehicle-mounted data acquisition and control subsystem, process the first driving data and the first road condition data to obtain second driving data and second road condition data, and upload the second driving data and the second road condition data to the cloud monitoring management platform;

[0008] The traffic flow prediction management subsystem is used to collect traffic flow data on the road based on video surveillance of the road intersection, and use the wavelet neural network to predict the predicted road traffic flow number for the next time period based on the traffic flow data of the previous three time periods, upload the traffic flow data and the predicted traffic flow number to the cloud monitoring management platform, and receive a data acceptance success signal from the cloud monitoring management platform;

[0009] The user management subsystem is used to collect the user's travel trajectory and travel behavior, and combine the user's historical data to obtain the user's next travel state transition matrix, and upload the next travel state transition matrix to the cloud monitoring management platform;

[0010] The cloud monitoring management platform is configured to receive and display the second driving data and the second road condition data sent by the road network data management subsystem, the traffic flow data and the predicted traffic flow data sent by the traffic flow prediction management subsystem, and the next travel state transfer matrix sent by the user management subsystem, process the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data, and the next travel state transfer matrix according to the electric vehicle charging load prediction method, calculate and display charging load prediction information when the electric vehicle is charging, the charging load prediction information including a time-scale load prediction curve and a space-scale load prediction curve, obtain a user charging path selection instruction based on the time-scale load prediction curve and the space-scale load prediction curve, and send the user charging path selection instruction to the user management subsystem;

[0011] The user management subsystem is further configured to receive the user charging path selection instruction sent by the cloud monitoring management platform, and guide the electric vehicle user to select a charging path according to the user charging path selection instruction.

[0012] Optionally, the cloud monitoring management platform includes an electric vehicle charging load prediction module, a cloud monitoring communication module, a cloud monitoring database and a cloud monitoring interface display module;

[0013] The cloud monitoring communication module is configured to receive the second driving data and the second road condition data sent by the road network data management subsystem, the traffic flow data and predicted traffic flow data sent by the traffic flow prediction management subsystem, and the next travel state transfer matrix sent by the user management subsystem;

[0014] The electric vehicle charging load prediction module is configured to calculate and display charging load prediction information for the electric vehicle when charging based on the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data, and the next travel state transfer matrix; the charging load prediction information includes: obtaining a prediction curve on a time scale and a prediction curve on a spatial scale based on the electric vehicle charging load prediction information, and obtaining a user charging path selection instruction based on the time scale load prediction curve and the spatial scale load prediction curve;

[0015] The cloud monitoring interface display module is configured to display the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data, the next travel state transfer matrix, and the charging load prediction information;

[0016] The cloud monitoring communication module is further used to send the user charging path selection instruction to the user management subsystem.

[0017] Optionally, the traffic flow prediction management subsystem includes a road video monitoring module, a traffic communication module, a traffic flow prediction module, and a traffic interface display module;

[0018] The road video surveillance module is used to collect traffic flow data within the first three Δt moments based on the video surveillance installed on the road;

[0019] The traffic communication module is used to send the traffic flow data within the first three Δt moments to the traffic flow prediction module;

[0020] The traffic flow prediction module is used to predict the road traffic flow number of the next Δt based on the traffic flow data in the previous three Δt moments using a wavelet neural network;

[0021] The traffic communication module is further used to send the traffic flow data within the first three Δt moments and the predicted road traffic flow number of the next Δt in the traffic flow prediction module to the cloud monitoring management platform.

[0022] Optionally, the vehicle-mounted data acquisition and control subsystem includes an infrared detection module, a driving data acquisition module, a data processing and control module, and a communication module;

[0023] The infrared detection module is used to detect the number of people currently in the electric vehicle cabin;

[0024] The driving data acquisition module is used to collect the driving data and road condition data of the electric vehicle, wherein the driving data includes the current position, the remaining power of the power battery, the operating status and operating parameters of the electric vehicle, the vehicle position, and the vehicle speed;

[0025] The data processing and control module is used to calculate the first driving data based on the current position, the remaining power of the power battery, the operating state and operating parameters of the electric vehicle, the vehicle position, the vehicle speed and the road condition data, wherein the first driving data includes the work done by the electric vehicle per unit time;

[0026] The communication module uploads the driving data, the road condition data and the first driving data to the road network data management subsystem, and then transmits the second driving data and the second road condition data processed by the road network data management subsystem to the cloud monitoring and scheduling platform.

[0027] Optionally, the system further includes a plurality of first cameras arranged in the vehicle, a plurality of second cameras arranged in the vehicle, and a plurality of video surveillance cameras arranged at road intersections; the first cameras are communicatively connected to the vehicle-mounted data processing subsystem via a CAN bus, and the second cameras are communicatively connected to the vehicle-mounted data processing subsystem via a CAN bus; the video surveillance cameras are communicatively connected to the vehicle flow monitoring subsystem via 5G communication technology;

[0028] The first camera is used to obtain the owner's facial expression history data, and the first camera is set on the triangular column of the car; the second camera is used to obtain the owner's head angle history data, and the second camera is set on the instrument panel assembly of the car; the video surveillance is used to collect the traffic video data on the road;

[0029] The vehicle-mounted data processing subsystem is also used to obtain several odometer remaining power data in the vehicle, several owner's expression history data sent by the first camera, and several owner's head angle history data sent by the first camera, and send the owner's expression history data, owner's head angle history data and odometer remaining power data to the cloud monitoring platform.

[0030] Optionally, the road network data management subsystem includes a data redundancy information processing module, a data normalization processing module, and a road network data storage module;

[0031] The data redundancy information processing module uses big data processing technology to process the collected first driving data and the first road condition data, process redundant and useless data information, and transmit the processed second driving data and the second road condition data to the data normalization processing module;

[0032] The data normalization processing module adopts a data normalization technology to perform evolutionary normalization processing on the first driving data and the first road condition data to obtain the second driving data and the second road condition data;

[0033] The road network data storage module is used to store the second driving data and the second road condition data in the data storage module.

[0034] Optionally, the user management subsystem includes a user data collection module, a user database, a user communication module, and a user data processing module;

[0035] The user data collection module is used to collect the user's travel trajectory and travel behavior, wherein the travel trajectory includes travel time, travel location and travel destination, and save the travel trajectory and travel behavior to the database;

[0036] The user data processing module calculates the next travel state transfer matrix of the electric vehicle user based on the travel trajectory, travel behavior and collected historical data; the next travel state transfer matrix is ​​the next travel destination state transfer matrix;

[0037] The user communication module is used to send the next travel state transfer matrix to the cloud monitoring management platform to provide user status data for charging load prediction of electric vehicles.

[0038] In addition, to achieve the above-mentioned purpose, the present invention provides a method for predicting the charging load of an electric vehicle based on traffic flow. The method for predicting the charging load of an electric vehicle based on traffic flow is applied to a system for predicting the charging load of an electric vehicle based on traffic flow. The method for predicting the charging load of an electric vehicle based on traffic flow comprises the following steps:

[0039] Step S10, using the traffic flow prediction management subsystem to process the traffic flow data of the first three time periods using a wavelet neural network to predict the traffic flow on the road within the next time Δt;

[0040] Step S20, obtaining the state of the traffic light at the intersection, and establishing an electric vehicle road travel impedance model according to the state of the traffic light at the intersection, and obtaining a total travel time model according to the electric vehicle road travel impedance model;

[0041] Step S30, constructing an adjacency matrix according to the road network graph, and performing dynamic path planning for the electric vehicle according to the adjacency matrix to obtain a shortest time cost matrix;

[0042] Step S40, predicting the charging load of the electric vehicle based on the predicted traffic flow on the road within the next time Δt, the total travel time model, and the shortest time cost matrix to obtain charging load prediction information;

[0043] The specific steps of step S10 are as follows:

[0044] Step S101 initializes the network parameters of the wavelet neural network, inputs the traffic flow data of the first three periods as samples into the wavelet neural network, and sets q(1,2,....,n) as the sample of the input signal. The input value of the pth sample is is the network output value of the p-th sample, is the output target value of the sample, and the connection weights between the output layer and the hidden layer, and between the hidden layer and the output layer are W ij 、W kj , let the scaling factor be a i , the translation factor is b i ,pass Shift and resize

[0045] Step S102: Create a wavelet function and select a mother wavelet function in the HiIbert vector space. Satisfy for The Fourier transform of The scaling and translation transformation of produces the wavelet function basis:

[0046]

[0047] Where a is the scaling factor and b is the translation factor; establish the activation function of the wavelet neural network;

[0048]

[0049] Among them, x i (i=1, 2...I) is the input signal of the i-th neuron in the input layer; y k (k=1, 2...k) is the output signal of the kth neuron in the output layer; w ji is the weight between the hidden layer neuron j and the output layer neuron i; is the weight between the input layer neuron j and the hidden layer node k; a j is the scaling factor of the jth hidden layer neuron; b j is the translation factor of the jth hidden layer neuron;

[0050] Step S103: predict the output of the network, calculate the error, and input a training sample (P k , T k ), k∈{1,2,...N}, where N is the number of training samples, P k is the sample input signal, T k Output target value for the network, P k ∈R m , T k ∈R m ; Calculate the prediction error:

[0051]

[0052] Step S104: Network weight correction, based on The parameter w of the wavelet neural network is modified by the steepest descent method. ji 、 a j 、b j ;

[0053]

[0054] Step S105, optimizing the convergence speed; in order to speed up the convergence speed of the network, a network parameter momentum factor α is introduced, so the iterative formula of the weight vector is:

[0055]

[0056] Step S106, determine whether the training is completed; after completing the above steps, determine Is it less than the required maximum error? If it is, the network training ends. If the number of network training times has reached the maximum number of training times specified, the training ends. Otherwise, the number of network training times is updated to t=t+1. Repeat the above steps.

[0057] Step S107: After steps S101, S102, S103, S104, S105 and S106, the predicted traffic flow number X on the road within the next moment Δt is output. predict .

[0058] Optionally, the step S20 is specifically as follows:

[0059] Step S201: Establish a road impedance model. Based on the classification of urban road traffic conditions, each type of traffic condition is divided into four categories: severe congestion, crowded, slow-moving, and smooth according to the map road conditions. When the saturation y is between 0 and 0.6, the traffic condition is smooth; when the saturation y is between 0.6 and 0.8, the traffic condition is slow-moving; when the saturation value y is between 0.8 and 1.0, the traffic condition is crowded; when the saturation value y is greater than 1.0, the traffic condition is severely congested. The road impedance model is:

[0060] T0=t0(1+α(y)) β 0≤y≤1

[0061] T0=t0(1+α(2-y)) β 1 <y≤2

[0062]

[0063] Where T0 is the impedance time, t0 is the travel time (s) when the traffic volume is zero; X predict The traffic volume of the actual road section is predicted, C is the road section capacity, y is the corrected saturation; η1 is the intersection interval influence correction coefficient, η2 is the non-motor vehicle interference influence correction coefficient; η3 is the pedestrian interference influence correction coefficient, η4 ​​is the lane width influence correction coefficient, α and β are impedance influence parameters;

[0064] Step S202: Establish a traffic node impedance model; the node impedance model is established according to the traffic light status at the intersection as follows:

[0065]

[0066] Where T1 is the node impedance time, c is the signal period, q is the vehicle arrival rate in the lane, t read is the red light time of the signal light at the node, λ is the green-to-signal ratio;

[0067] Step S203: According to S201 and S202, the overall impedance model is obtained as follows:

[0068]

[0069] Step S204: The free flow travel time model is obtained according to the speed in the free flow state of the road:

[0070]

[0071] Where T free is the free flow state travel time, L is the length of the road section, v free is the free flow speed, which is generally the maximum speed for which the road is designed;

[0072] Step S205: According to step S203 and step S204, the total time model of the passage is obtained as follows:

[0073] T total =T+T free .

[0074] Optionally, the step S30 is specifically as follows:

[0075] Step S301: Initialize the adjacency matrix w of the road network graph (0) , the time cost between i and j is T total =T+T free Indicates that if points i and j are not connected, then T total =∞;

[0076] Step S302, construct w (1) , insert the first intermediate node between i and j, and calculate

[0077] Step S303, construct w (2) , insert a second intermediate node between i and j, and calculate

[0078] Step S304, according to step S301, step S302, step S303, construct w (n) ,calculate Then w (n) is the shortest path length between i and j after traversing all nodes, w (n) It is the shortest time cost matrix between each point.

[0079] Optionally, the step S40 is specifically as follows:

[0080] Step S401, calculating the number of electric vehicles on the road based on the traffic flow prediction data and the predicted number of vehicles in the traffic flow;

[0081] Among them, the number of large electric vehicles N EV The method for obtaining (big) includes the following steps:

[0082] Collect traffic video data on the road based on video surveillance at road intersections;

[0083] The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a large electric vehicle, thereby determining the number of large electric vehicles on the road.

[0084] The number N of small electric vehicles EV(small) acquisition method, the specific steps include:

[0085] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0086] The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a small electric vehicle, thereby determining the number of small electric vehicles on the road.

[0087] The number of electric vehicles N EV The specific steps of obtaining include:

[0088] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0089] The YOLO v4 algorithm is used to monitor the traffic video data in real time, and the color and length of the license plate of the electric vehicle are identified according to the image recognition algorithm to determine whether the vehicle is an electric vehicle, and then the number of electric vehicles on the road is determined;

[0090] The number of electric vehicles with charging needs The specific steps of obtaining include:

[0091] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0092] The traffic video data is read to obtain the facial expression data and head angle data of the vehicle owner traveling on the monitored road;

[0093] Input the owner's facial expression data and head angle data into the vehicle's remaining power model to obtain the number of electric vehicles that need charging;

[0094] The specific steps of training the vehicle remaining power model are:

[0095] The vehicle data acquisition and control subsystem obtains the owner's facial expression history data collected by the first camera in the vehicle, the owner's head angle history data collected by the second camera, and the odometer remaining power data;

[0096] The owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are processed through time series, and then the processed owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are input into the BP neural network for training to calculate the vehicle remaining power model;

[0097] Step S402: The current state of charge of the electric vehicle is determined. If the current state of charge is less than 30% of the capacity, the user management subsystem sends a charging instruction to the cloud monitoring management platform. At the same time, the user management subsystem counts the number and location of electric vehicles that need to be charged and transmits it to the cloud monitoring management platform. The electric vehicle charging demand charging model is:

[0098]

[0099] in is the state of charge of the electric vehicle at the start, C x is the battery capacity;

[0100] The location model of electric vehicles in the road network is:

[0101]

[0102] in is the initial position of the electric vehicle in the road network, X i,j is the location of the transportation node, X i+1,j is the next traffic node location horizontally, X i,j+1 is the location of the longitudinal traffic node;

[0103] Step S403: Based on the impedance model data, combined with steps S10 and S20, the state of charge of the electric vehicle when it arrives at the charging station is obtained.

[0104]

[0105] in is the state of charge of the electric vehicle when it arrives at the charging station, and P is the work done per unit time by the electric vehicle while driving on the road;

[0106] The time model for reaching the charging station is:

[0107]

[0108] in is the time to reach the charging station, The time when charging demand occurs;

[0109] Step S404: According to step S10, step S20, and step S30, the electric vehicle charging model is obtained as follows:

[0110]

[0111] in Charging time of large vehicles with fast charge, SOC max Charging maximum state of charge, Charging time for large vehicles with slow charging, Charging time for small vehicles with fast charging, Charging time for small vehicles with slow charging, P fast is the fast charging power, P slow is the slow charge power;

[0112] Step S405, combining steps S10, S20, and S30 to obtain a charging load prediction model:

[0113] P(T) Total =P(T) big +P(T) small

[0114] Where P(T) big Charging power for large vehicles, P(T) small Charging power for small cars; When slow charging is selected, it is 1, and when not selected, it is 0; When fast charging is selected, it is 1, and when not selected, it is 0; P(T) Total is the total charging power at time T; The number of large vehicles that charge at charging stations, The number of small cars that charge at charging stations;

[0115] Step S406, charging station location model;

[0116]

[0117]

[0118] in is the initial position of the electric vehicle in the road network, X i,j is the location of the transportation node, X i+1,j is the next traffic node location horizontally, X i,j+1 is the location of the longitudinal traffic node;

[0119] Step S407, combining step S401, step S402, step S403, step S404, step S405, and step S406, obtains a charging load prediction curve of electric vehicles within a day, with a time interval of Δt. At the same time, a charging load prediction curve of each charging station within a day can also be obtained spatially.

[0120] The beneficial effects of the present invention are as follows:

[0121] 1. The present invention uses a method based on data regularity to predict the charging load of regional charging stations, which can avoid the influence of uncertain data such as user charging regularity and battery charging characteristics, and improve the accuracy and applicability of charging load prediction.

[0122] 2. The present invention can accurately predict the load of regional electric vehicle charging facilities, providing a basis for studying the impact of electric vehicle charging facility load on the power grid and also providing a basis for electric vehicle charging facility planning.

[0123] 3. The present invention conducts comprehensive data mining on various factors that can affect the charging load of electric vehicles, quantitatively analyzes the impact of various factors on the charging load of electric vehicles, and makes the prediction model of electric vehicle charging load more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0124] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the proportional relationships of the various components in the drawings of this specification do not represent the proportional relationships in actual material selection and design, and are merely schematic diagrams of structures or positions, among which:

[0125] Figure 1 This is an architecture diagram of the electric vehicle charging load prediction system based on traffic flow of the present invention;

[0126] Figure 2 It is the wavelet neural network prediction flow chart of the present invention;

[0127] Figure 3 It is the overall flow chart of the present invention;

[0128] Figure 4 The road topology of the present invention, electric vehicles and charging stations are value graphs. DETAILED DESCRIPTION

[0129] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. That is, the embodiments described are only some embodiments of the present invention, rather than all embodiments.

[0130] Example 1

[0131] The technical solution adopted in the present invention is as follows:

[0132] like Figure 1 、 Figure 4 As shown in the figure, a traffic flow-based electric vehicle charging load forecasting system consists of a management platform and four subsystems:

[0133] Cloud monitoring and management platform: Receives and displays relevant data and operating status from the road network data management subsystem, traffic flow forecasting and management subsystem, vehicle data acquisition and control subsystem, and user management subsystem in real time. It processes this data to derive spatiotemporal load forecast information for electric vehicle charging and transmits it to the load forecast display interface, which displays the load situation on both temporal and spatial scales. It also monitors the operation of each subsystem in real time and issues an alarm for timely maintenance in the event of any abnormality in travel.

[0134] On-board data acquisition and control subsystem: collects real-time traffic data changes of electric vehicles and uploads the processed driving data and road condition data to the road network data management subsystem;

[0135] Road network data management subsystem: receives driving data and road condition data from the vehicle-mounted data acquisition and control subsystem, processes the data, and uploads the processed data to the cloud monitoring management platform;

[0136] Traffic flow prediction and management subsystem: Traffic flow data on the road is collected based on video surveillance at road intersections. The data is uploaded to the traffic flow prediction and management subsystem. The wavelet neural network is used to predict the traffic flow of the next period based on the traffic flow data of the previous moment. After processing, the processed data is uploaded to the cloud monitoring management platform, and a data acceptance success signal is received from the cloud monitoring management platform, indicating that the data has been successfully received.

[0137] User management subsystem: collects users' travel trajectories and records their travel behaviors. Combined with historical data, it obtains the user's travel status transition probability model for the next day. The processed change data is uploaded to the cloud monitoring management platform, and the system receives instructions from the cloud monitoring management platform to guide electric vehicle users in choosing charging paths.

[0138] Furthermore, the cloud monitoring management platform includes an electric vehicle charging load prediction module, a communication module, a database, and an interface display module;

[0139] The electric vehicle charging load prediction module predicts changes in electric vehicle load based on collected data related to electric vehicles, traffic flow, and user travel. The electric vehicle charging load prediction includes: combining the electric vehicle load prediction curve to obtain a prediction curve on a time scale and a prediction curve on a spatial scale, so as to achieve real-time and accurate prediction of the charging load of electric vehicle clusters.

[0140] Furthermore, the traffic flow prediction management subsystem includes a road video monitoring module, a communication module, a traffic flow prediction module, and an interface display module;

[0141] The road video surveillance module collects traffic flow data within the first three Δt moments based on the video surveillance installed on the road;

[0142] The traffic flow prediction module receives data from the road video monitoring module through the transmission of the communication module, and uses the wavelet neural network to predict the road traffic flow number of the next Δt, so as to achieve the purpose of real-time prediction of the traffic flow at the next moment.

[0143] Furthermore, the vehicle-mounted data acquisition and control subsystem includes an infrared detection module, a driving data acquisition module, a data processing and control module, and a communication module;

[0144] The infrared detection module detects the number of people currently in the electric vehicle cabin;

[0145] The driving data acquisition module collects the current position of the electric vehicle, the remaining power of the power battery, the operating status and operating parameters of the electric vehicle, the vehicle position, speed, and road condition information, and calculates the work done per unit time by the electric vehicle;

[0146] The communication module uses a 5G network to upload the electric vehicle data collected by the driving data collection module to the road network data management subsystem, and then transmits the data to the cloud monitoring and scheduling platform after being processed by the road network data management subsystem.

[0147] The electric vehicle charging load prediction system based on traffic flow also includes a plurality of first cameras arranged in the vehicle, a plurality of second cameras arranged in the vehicle, and a plurality of video surveillance cameras arranged at road intersections; the first cameras are communicatively connected to the vehicle data processing subsystem via a CAN bus, and the second cameras are communicatively connected to the vehicle data processing subsystem via a CAN bus; the video surveillance cameras are communicatively connected to the vehicle flow monitoring subsystem via 5G communication technology;

[0148] The first camera is used to obtain the owner's facial expression history data, and the first camera is set on the triangular column of the car; the second camera is used to obtain the owner's head angle history data, and the second camera is set on the instrument panel assembly of the car; the video surveillance is used to collect the traffic video data on the road;

[0149] The vehicle-mounted data processing subsystem is also used to obtain several odometer remaining power data in the vehicle, several owner's expression history data sent by the first camera, and several owner's head angle history data sent by the first camera, and send the owner's expression history data, owner's head angle history data and odometer remaining power data to the cloud monitoring platform.

[0150] Furthermore, the road network data management subsystem includes a data redundancy information processing module, a data normalization processing module, and a data storage module;

[0151] The data redundancy information processing module adopts big data processing technology to process various collected driving data, process redundant and useless data information, and transmit the processed data to the data normalization processing module;

[0152] The data normalization processing module adopts data normalization technology to normalize the processed driving data and unify the data to facilitate subsequent calculations;

[0153] The data processing module stores the processed information in the data storage module to facilitate data transmission and reception. The user management subsystem includes a user data collection module, a database, a communication module, and a user data processing module. The user data collection module collects the user's travel time, location, and destination and saves the collected data to the database. The user data processing module combines the collected historical data to calculate the destination state transition matrix of the electric vehicle user's next trip.

[0154] The communication module transmits the data processed by the user data processing module to the cloud monitoring management platform to provide user status data for charging load prediction of electric vehicles.

[0155] like Figure 2 、 Figure 3 As shown, a method for predicting electric vehicle charging load based on traffic flow includes the following steps:

[0156] Step S10, performing traffic flow prediction;

[0157] Step S20, establishing an electric vehicle road driving impedance model;

[0158] Step S30, performing dynamic path planning for the electric vehicle;

[0159] Step S40, predicting the charging load of the electric vehicle;

[0160] In step S10, traffic flow prediction is performed using wavelet function modeling. A wavelet neural network is a new type of neural network developed based on the BP neural network, using the wavelet basis function as the activation function of the hidden layer. The main difference between this type of neural network and the BP neural network is the different activation function used. Therefore, like the BP neural network, it is a feedforward neural network with feedback. The wavelet transform enables the network to learn quickly while also preventing data from falling into local minima. This results in faster convergence and higher prediction accuracy than the BP neural network.

[0161] Step S101 first initializes the relevant parameters, let q(1,2,....,n) be the sample of the input signal, and the input value of the pth sample is is the network output value of the p-th sample, is the output target value of the sample, and the connection weights between the output layer and the hidden layer, and between the hidden layer and the output layer are W ij 、W kj , let the scaling factor be a i , the translation factor is b i , translate and scale according to the following formula,

[0162]

[0163] Step S102: Create a wavelet function and select a mother wavelet function in the HiIbert vector space. Satisfy for The Fourier transform of The scaling and translation transformation of produces the wavelet function basis:

[0164]

[0165] Where a and b are scaling factors and translation factors respectively; establish the activation function of wavelet neural network;

[0166]

[0167] Among them, x i (i=1, 2...I) is the input signal of the i-th neuron in the input layer; y k (k=1, 2...k) is the output signal of the kth neuron in the output layer; w ji is the weight between the hidden layer neuron j and the output layer neuron i; is the weight between the input layer neuron j and the hidden layer node k; a j is the scaling factor of the jth hidden layer neuron; b j is the translation factor of the jth hidden layer neuron;

[0168] Step S103: predict the output of the network, calculate the error, and input a training sample (P k , T k ), k∈{1,2,...N}, where N is the number of training samples, P k is the sample input signal, T k Output target value for the network, P k ∈R m , T k ∈R m ; Calculate the prediction error:

[0169]

[0170] Step S104: Network weight correction, based on The parameter w of the wavelet neural network is modified by the steepest descent method. ji 、 a j 、b j ;

[0171]

[0172] Step S105, optimizing the convergence speed; in order to speed up the convergence speed of the network, a network parameter momentum factor α is introduced, so the iterative formula of the weight vector is:

[0173]

[0174] Step S106, determine whether the training is completed; after completing the above steps, determine Is it less than the required maximum error? If it is, the network training ends. If the number of network training times has reached the maximum number of training times specified, the training ends. Otherwise, the number of network training times is updated to t=t+1. Repeat the above steps.

[0175] Step S107: After steps S101, S102, S103, S104, S105 and S106, the predicted traffic flow number X on the road within the next moment Δt is output. predict .

[0176] The specific steps of the electric vehicle road driving impedance model in step S20 are as follows:

[0177] Step S201: Establish a road impedance model. Based on the classification of urban road traffic conditions, each type of traffic condition is divided into four categories: severe congestion, crowded, slow-moving, and smooth according to the map road conditions. When the saturation y is between 0 and 0.6, the traffic condition is smooth; when the saturation y is between 0.6 and 0.8, the traffic condition is slow-moving; when the saturation value y is between 0.8 and 1.0, the traffic condition is crowded; when the saturation value y is greater than 1.0, the traffic condition is severely congested. The road impedance model is:

[0178] T0=t0(1+α(y)) β 0≤y≤1

[0179] T0=t0(1+α(2-y)) β 1 <y≤2

[0180]

[0181] Where T0 is the impedance time, t0 is the travel time (s) when the traffic volume is zero; X predict The traffic volume of the actual road section is predicted, C is the road section capacity, y is the corrected saturation; η1 is the intersection interval influence correction coefficient, η2 is the non-motor vehicle interference influence correction coefficient; η3 is the pedestrian interference influence correction coefficient, η4 ​​is the lane width influence correction coefficient, α and β are impedance influence parameters;

[0182] Step S202: Establish a traffic node impedance model; the node impedance model is established according to the traffic light status at the intersection as follows:

[0183]

[0184] Where T1 is the node impedance time, c is the signal period, q is the vehicle arrival rate in the lane, t read is the red light time of the signal light at the node, λ is the green-to-signal ratio;

[0185] Step S203: According to S201 and S202, the overall impedance model is obtained as follows:

[0186]

[0187]

[0188] Step S204: The free flow travel time model is obtained according to the speed in the free flow state of the road:

[0189]

[0190] Where T free is the free flow state travel time, L is the length of the road section, v free is the free flow speed, which is generally the maximum speed for which the road is designed;

[0191] Step S205: According to step S203 and step S204, the total time model of the passage is obtained as follows:

[0192] T total =T+T free .

[0193] The specific steps of step S30, dynamic path planning of electric vehicles are as follows:

[0194] The Floyd shortest path algorithm, also known as the insertion point method, is a dynamic programming algorithm with simple calculations. It can solve the shortest path between multiple source points and is an effective method for solving the shortest path problem between any two points. The main idea of ​​the Floyd algorithm is to start with the weighted adjacency matrix of any two points i and j, assign a value to the length between any two points, and if the two points are not connected, it is represented by infinity inf. Each time a new node k is inserted, the known length between points i and j is compared with the path length with node k as the intermediate transition point. The smaller value of the comparison result is used as the new distance matrix. The new node is then inserted repeatedly n times to obtain n distance matrices, which are recorded as D(1), D(2), D(3)...D(n). Then D(n) is the shortest path distance information between each vertex in the graph.

[0195] Step S301: Initialize the adjacency matrix w of the road network graph (0) , the time cost between i and j is T total =T+T free Indicates that if points i and j are not connected, then T total =∞;

[0196] Step S302, construct w (1) , insert the first intermediate node between i and j, and calculate

[0197] Step S303, construct w (2) , insert a second intermediate node between i and j, and calculate

[0198] Step S304, according to step S301, step S302, step S303, construct w (n) ,calculate Then w (n) is the shortest path length between i and j after traversing all nodes, w (n) It is the shortest time cost matrix between each point.

[0199] The specific steps of the step S40, electric vehicle charging load prediction, are as follows:

[0200] Step S401, calculating the number of electric vehicles on the road based on the traffic flow prediction data and the predicted number of vehicles in the traffic flow;

[0201] Among them, the number of large electric vehicles N EV The method for obtaining (big) includes the following steps:

[0202] Collect traffic video data on the road based on video surveillance at road intersections;

[0203] The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a large electric vehicle, thereby determining the number of large electric vehicles on the road.

[0204] The number N of small electric vehicles EV (small) acquisition method, the specific steps include:

[0205] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0206] The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a small electric vehicle, thereby determining the number of small electric vehicles on the road.

[0207] The number of electric vehicles N EV The specific steps of obtaining include:

[0208] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0209] The YOLO v4 algorithm is used to monitor the traffic video data in real time, and the color and length of the license plate of the electric vehicle are identified according to the image recognition algorithm to determine whether the vehicle is an electric vehicle, and then the number of electric vehicles on the road is determined;

[0210] The number of electric vehicles with charging needs The specific steps of obtaining include:

[0211] Collecting the traffic video data on the road based on video surveillance at the road intersection;

[0212] The traffic video data is read to obtain the facial expression data and head angle data of the vehicle owner traveling on the monitored road;

[0213] Input the owner's facial expression data and head angle data into the vehicle's remaining power model to obtain the number of electric vehicles that need charging;

[0214] The specific steps of training the vehicle remaining power model are:

[0215] The vehicle data acquisition and control subsystem obtains the owner's facial expression history data collected by the first camera in the vehicle, the owner's head angle history data collected by the second camera, and the odometer remaining power data;

[0216] The owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are processed through time series, and then the processed owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are input into the BP neural network for training to calculate the vehicle remaining power model;

[0217] Step S402: The current state of charge of the electric vehicle is determined. If the current state of charge is less than 30% of the capacity, the user management subsystem sends a charging instruction to the cloud monitoring management platform. At the same time, the user management subsystem counts the number and location of electric vehicles that need to be charged and transmits it to the cloud monitoring management platform. The electric vehicle charging demand charging model is:

[0218]

[0219] in is the state of charge of the electric vehicle at the start, C x is the battery capacity;

[0220] The location model of electric vehicles in the road network is:

[0221]

[0222] in is the initial position of the electric vehicle in the road network, X i,j is the location of the transportation node, X i+1,j is the next traffic node location horizontally, X i,j+1 is the location of the longitudinal traffic node;

[0223] Step S403: Based on the impedance model data, combined with steps S10 and S20, the state of charge of the electric vehicle when it arrives at the charging station is obtained.

[0224]

[0225] in is the state of charge of the electric vehicle when it arrives at the charging station, and P is the work done per unit time by the electric vehicle while driving on the road;

[0226] The time model for reaching the charging station is:

[0227]

[0228] in is the time to reach the charging station, The time when charging demand occurs;

[0229] Step S404: According to step S10, step S20, and step S30, the electric vehicle charging model is obtained as follows:

[0230]

[0231] in Charging time of large vehicles with fast charge, SOC max Charging maximum state of charge, Charging time for large vehicles with slow charging, Charging time for small vehicles with fast charging, Charging time for small vehicles with slow charging, P fast is the fast charging power, P slow is the slow charge power;

[0232] Step S405, combining steps S10, S20, and S30 to obtain a charging load prediction model:

[0233] P(T) Total =P(T) big +P(T) small

[0234] Where P(T) big Charging power for large vehicles, P(T) small Charging power for small cars; When slow charging is selected, it is 1, and when not selected, it is 0; When fast charging is selected, it is 1, and when not selected, it is 0; P(T) Total is the total charging power at time T; The number of large vehicles that charge at charging stations, The number of small cars that charge at charging stations;

[0235] Step S406, charging station location model;

[0236]

[0237] in is the initial position of the electric vehicle in the road network, X i,j is the location of the transportation node, X i+1,j is the next traffic node location horizontally, X i,j+1 is the location of the longitudinal traffic node;

[0238] Step S407, combined with step S401, step S402, step S403, step S404, step S405, and step S406, can obtain a charging load prediction curve of electric vehicles within a day, with a time interval of Δt. At the same time, a charging load prediction curve of each charging station within a day can also be obtained spatially.

[0239] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A system for predicting electric vehicle charging load based on traffic flow, characterized in that: The electric vehicle charging load prediction system includes a cloud monitoring management platform, a road network data management subsystem, a traffic flow prediction management subsystem, an on-board data acquisition and control subsystem, and a user management subsystem; The vehicle-mounted data acquisition and control subsystem is used to collect driving data and road condition data of the electric vehicle participating in real-time traffic, process the driving data and road condition data to obtain first driving data and first road condition data, and upload the first driving data and the first road condition data to the road network data management subsystem; The road network data management subsystem is configured to receive the first driving data and the first road condition data sent by the vehicle-mounted data acquisition and control subsystem, process the first driving data and the first road condition data to obtain second driving data and second road condition data, and upload the second driving data and the second road condition data to the cloud monitoring management platform; The traffic flow prediction management subsystem is used to collect traffic flow data on the road based on video surveillance of the road intersection, and use the wavelet neural network to predict the predicted road traffic flow number for the next time period based on the traffic flow data of the previous three time periods, upload the traffic flow data and the predicted traffic flow number to the cloud monitoring management platform, and receive a data acceptance success signal from the cloud monitoring management platform; The user management subsystem is used to collect the user's travel trajectory and travel behavior, and combine the user's historical data to obtain the user's next travel state transition matrix, and upload the next travel state transition matrix to the cloud monitoring management platform; The cloud monitoring management platform is configured to receive and display the second driving data and the second road condition data sent by the road network data management subsystem, the traffic flow data and the predicted traffic flow data sent by the traffic flow prediction management subsystem, and the next travel state transfer matrix sent by the user management subsystem, process the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data, and the next travel state transfer matrix according to the electric vehicle charging load prediction method, calculate and display charging load prediction information when the electric vehicle is charging, the charging load prediction information including a time-scale load prediction curve and a space-scale load prediction curve, obtain a user charging path selection instruction based on the time-scale load prediction curve and the space-scale load prediction curve, and send the user charging path selection instruction to the user management subsystem; The user management subsystem is further configured to receive the user charging path selection instruction sent by the cloud monitoring management platform, and guide the electric vehicle user to select a charging path according to the user charging path selection instruction.

2. The electric vehicle charging load prediction system based on traffic flow according to claim 1 is characterized in that: The cloud monitoring management platform includes an electric vehicle charging load prediction module, a cloud monitoring communication module, a cloud monitoring database and a cloud monitoring interface display module; The cloud monitoring communication module is configured to receive the second driving data and the second road condition data sent by the road network data management subsystem, the traffic flow data and predicted traffic flow data sent by the traffic flow prediction management subsystem, and the next travel state transfer matrix sent by the user management subsystem; The electric vehicle charging load prediction module is used to calculate and display charging load prediction information when the electric vehicle is charging based on the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data and the next travel state transfer matrix; The charging load prediction information includes: obtaining a prediction curve on a time scale and a prediction curve on a space scale according to the electric vehicle charging load prediction information, and obtaining a user charging path selection instruction according to the time scale load prediction curve and the space scale load prediction curve; The cloud monitoring interface display module is configured to display the second driving data, the second road condition data, the traffic flow data, the predicted traffic flow data, the next travel state transfer matrix, and the charging load prediction information; The cloud monitoring communication module is further used to send the user charging path selection instruction to the user management subsystem.

3. The electric vehicle charging load prediction system based on traffic flow according to claim 1 is characterized in that: The traffic flow prediction management subsystem includes a road video monitoring module, a traffic communication module, a traffic flow prediction module, and a traffic interface display module; The road video surveillance module is used to collect traffic flow data within the first three Δt moments based on the video surveillance installed on the road; The traffic communication module is used to send the traffic flow data within the first three Δt moments to the traffic flow prediction module; The traffic flow prediction module is used to predict the road traffic flow number of the next Δt based on the traffic flow data in the previous three Δt moments using a wavelet neural network; The traffic communication module is further used to send the traffic flow data within the first three Δt moments and the predicted road traffic flow number of the next Δt in the traffic flow prediction module to the cloud monitoring management platform.

4. The electric vehicle charging load prediction system based on traffic flow according to claim 1 is characterized in that: The vehicle-mounted data acquisition and control subsystem includes an infrared detection module, a driving data acquisition module, a data processing and control module, and a communication module; The infrared detection module is used to detect the number of people currently in the electric vehicle cabin; The driving data acquisition module is used to collect the driving data and road condition data of the electric vehicle, wherein the driving data includes the current position, the remaining power of the power battery, the operating status and operating parameters of the electric vehicle, the vehicle position, and the vehicle speed; The data processing and control module is used to calculate the first driving data based on the current position, the remaining power of the power battery, the operating state and operating parameters of the electric vehicle, the vehicle position, the vehicle speed and the road condition data, wherein the first driving data includes the work done by the electric vehicle per unit time; The communication module is used to upload the driving data, the road condition data and the first driving data to the road network data management subsystem, and then transmit the second driving data and the second road condition data processed by the road network data management subsystem to the cloud monitoring management platform.

5. The electric vehicle charging load prediction system based on traffic flow according to claim 4 is characterized in that: The system also includes a plurality of first cameras arranged in the vehicle, a plurality of second cameras arranged in the vehicle, and a plurality of video surveillance cameras arranged at road intersections; the first cameras are communicatively connected to the vehicle-mounted data acquisition and control subsystem via a CAN bus, and the second cameras are communicatively connected to the vehicle-mounted data acquisition and control subsystem via a CAN bus; the video surveillance cameras are communicatively connected to the traffic flow prediction and management subsystem via 5G communication technology; The first camera is used to obtain the owner's facial expression history data, and the first camera is set on the car's triangular column; the second camera is used to obtain the owner's head angle history data, and the second camera is set on the car's instrument panel assembly; the video surveillance is used to collect traffic video data on the road; The vehicle-mounted data acquisition and control subsystem is also used to obtain several odometer remaining power data in the vehicle, several owner's expression history data sent by the first camera, and several owner's head angle history data sent by the first camera, and send the owner's expression history data, owner's head angle history data and odometer remaining power data to the cloud monitoring management platform.

6. The electric vehicle charging load prediction system based on traffic flow according to claim 1 is characterized in that: The road network data management subsystem includes a data redundancy information processing module, a data normalization processing module, and a road network data storage module; The data redundancy information processing module uses big data processing technology to process the collected first driving data and the first road condition data, process redundant and useless data information, and transmit the processed second driving data and the second road condition data to the data normalization processing module; The data normalization processing module adopts a data normalization technology to perform evolutionary normalization processing on the first driving data and the first road condition data to obtain the second driving data and the second road condition data; The road network data storage module is used to store the second driving data and the second road condition data in the data storage module; The user management subsystem includes a user data acquisition module, a user database, a user communication module, and a user data processing module; The user data collection module is used to collect the user's travel trajectory and travel behavior, wherein the travel trajectory includes travel time, travel location and travel destination, and save the travel trajectory and travel behavior to the database; The user data processing module is used to calculate the next travel state transfer matrix of the electric vehicle user based on the travel trajectory, travel behavior and collected historical data; the next travel state transfer matrix is ​​the next travel destination state transfer matrix; The user communication module is used to send the next travel state transfer matrix to the cloud monitoring management platform to provide user status data for charging load prediction of electric vehicles.

7. A method for predicting electric vehicle charging load based on traffic flow, characterized in that: The electric vehicle charging load prediction method is applied to the electric vehicle charging load prediction system according to any one of claims 1 to 6, comprising the following steps: Step S10, using the traffic flow prediction management subsystem to process the traffic flow data of the first three time periods using a wavelet neural network to predict the traffic flow on the road within the next time Δt; Step S20, obtaining the state of the traffic light at the intersection, and establishing an electric vehicle road travel impedance model according to the state of the traffic light at the intersection, and obtaining a total travel time model according to the electric vehicle road travel impedance model; Step S30, constructing an adjacency matrix according to the road network graph, and performing dynamic path planning for the electric vehicle according to the adjacency matrix to obtain a shortest time cost matrix; Step S40, predicting the charging load of the electric vehicle based on the predicted traffic flow on the road within the next time Δt, the total travel time model, and the shortest time cost matrix to obtain charging load prediction information; The specific steps of step S10 are as follows: Step S101 initializes the network parameters of the wavelet neural network, inputs the traffic flow data of the first three periods as samples into the wavelet neural network, and sets q as the sample of the input signal, and the input value of the pth sample is is the network output value of the p-th sample, is the output target value of the sample, and the connection weights between the output layer and the hidden layer, and between the hidden layer and the output layer are W ij 、W kj , let the scaling factor be a i , the translation factor is b i ,pass Perform translation and extension; Step S102: Create a wavelet function and select a mother wavelet function in the HiIbert vector space. Satisfy for The Fourier transform of The scaling and translation transformation of produces the wavelet function basis: Where a is the scaling factor and b is the translation factor; establish the activation function of the wavelet neural network; Among them, x i is the input signal of the i-th neuron in the input layer; y k is the output signal of the kth neuron in the output layer; w ji is the weight between the hidden layer neuron j and the output layer neuron i; is the weight between input layer neuron j and hidden layer node k; a j is the scaling factor of the jth hidden layer neuron; b j is the translation factor of the jth hidden layer neuron; Step S103: predict the output of the network, calculate the error, and input a training sample (P k , T k ), k∈{1,2,...N}, where N is the number of training samples, P k is the sample input signal, T k Output target value for the network, P k ∈R m , T k ∈R m ; Calculate the prediction error: Step S104: Network weight correction, based on The parameter w of the wavelet neural network is modified by the steepest descent method. ji 、 a j 、b j ; Step S105, optimizing the convergence speed; in order to speed up the convergence speed of the network, a network parameter momentum factor α is introduced, so the iterative formula of the weight vector is: Step S106, determine whether the training is completed; after completing the above steps, determine Is it less than the required maximum error? If it is, the network training ends. If the number of network training times has reached the maximum number of training times specified, the training ends. Otherwise, the number of network training times is updated to t=t+1. Repeat the above steps. Step S107: After steps S101, S102, S103, S104, S105 and S106, the predicted traffic flow number X on the road within the next moment Δt is output. predict .

8. The method for predicting electric vehicle charging load based on traffic flow according to claim 7, characterized in that: The step S20 is specifically as follows: Step S201: Establish a road impedance model. Based on the classification of urban road traffic conditions, each type of traffic condition is divided into four categories: severe congestion, crowded, slow-moving, and smooth according to the map road conditions. When the saturation y is between 0 and 0.6, the traffic condition is smooth; when the saturation y is between 0.6 and 0.8, the traffic condition is slow-moving; when the saturation value y is between 0.8 and 1.0, the traffic condition is crowded; when the saturation value y is greater than 1.0, the traffic condition is severe congestion. The road impedance model is: T0=t0(1+α(y)) β 0≤y≤1 T0=t0(1+α(2-y)) β 1<y≤2 Where T0 is the impedance time, and t0 is the travel time of the road section when the traffic volume is zero; X predict The traffic volume of the actual road section is predicted, C is the road section capacity, y is the corrected saturation; η1 is the intersection interval influence correction coefficient, η2 is the non-motor vehicle interference influence correction coefficient; η3 is the pedestrian interference correction coefficient, η4 ​​is the lane width correction coefficient, α and β are the impedance influence parameters; Step S202: Establish a traffic node impedance model; the node impedance model is established according to the traffic light status at the intersection as follows: Where T1 is the node impedance time, c is the signal period, q is the vehicle arrival rate in the lane, t read is the red light time of the signal light at the node, λ is the green-to-signal ratio; Step S203: According to S201 and S202, the overall impedance model is obtained as follows: Step S204: The free flow travel time model is obtained according to the speed in the free flow state of the road: Where T free is the free flow state travel time, L is the length of the road section, v free is the free flow speed, which indicates the maximum speed for which the road is designed; Step S205: According to step S203 and step S204, the total time model of the passage is obtained as follows: T total =T+T free 。 9. The method for predicting electric vehicle charging load based on traffic flow according to claim 8, characterized in that: The step S30 is specifically as follows: Step S301: Initialize the adjacency matrix w of the road network graph (0) , the time cost between i and j is Indicates that if points i and j are not connected, then T total =∞; Step S302, construct w (1) , insert the first intermediate node between i and j, and calculate Step S303, construct w (2) , insert a second intermediate node between i and j, and calculate Step S304, according to step S301, step S302, step S303, construct w (n) ,calculate Then w (n) is the shortest path length between i and j after traversing all nodes, w (n) It is the shortest time cost matrix between each point.

10. The method for predicting electric vehicle charging load based on traffic flow according to claim 9, characterized in that: The step S40 is specifically as follows: Step S401, calculating the number of electric vehicles on the road based on the traffic flow prediction data and the predicted number of vehicles in the traffic flow; Among them, the number of large electric vehicles N EV The method for obtaining (big) includes the following steps: Collect traffic video data on the road based on video surveillance at road intersections; The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a large electric vehicle, thereby determining the number of large electric vehicles on the road. Number of small electric vehicles N EV (small) acquisition method, the specific steps include: Collecting the traffic video data on the road based on video surveillance at the road intersection; The traffic video data is monitored in real time using the YOLO v4 algorithm, and the color, license plate number length, and vehicle size of the electric vehicle are identified using an image recognition algorithm to determine whether the vehicle is a small electric vehicle, thereby determining the number of small electric vehicles on the road. The number of electric vehicles N EV The specific steps of obtaining include: Collecting the traffic video data on the road based on video surveillance at the road intersection; The YOLO v4 algorithm is used to monitor the traffic video data in real time, and the color and length of the license plate of the electric vehicle are identified according to the image recognition algorithm to determine whether the vehicle is an electric vehicle, and then the number of electric vehicles on the road is determined; Number of electric vehicles with charging needs The specific steps of obtaining include: Collecting the traffic video data on the road based on video surveillance at the road intersection; The traffic video data is read to obtain the facial expression data and head angle data of the vehicle owner traveling on the monitored road; Input the owner's facial expression data and head angle data into the vehicle's remaining power model to obtain the number of electric vehicles that need charging; The specific steps for training the vehicle remaining power model are: The vehicle data acquisition and control subsystem obtains the owner's facial expression history data collected by the first camera in the vehicle, the owner's head angle history data collected by the second camera, and the odometer remaining power data; The owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are processed through time series, and then the processed owner's facial expression history data, the owner's head angle history data, and the odometer remaining power data are input into the BP neural network for training to calculate the vehicle remaining power model; Step S402: The current state of charge of the electric vehicle is determined. If the current state of charge is less than 30% of the capacity, the user management subsystem sends a charging instruction to the cloud monitoring management platform. At the same time, the user management subsystem counts the number and location of electric vehicles that need to be charged and transmits it to the cloud monitoring management platform. The electric vehicle charging demand charging model is: in is the state of charge of the electric vehicle at the start, C x is the battery capacity; The location model of electric vehicles in the road network is: in is the initial position of the electric vehicle in the road network, X i,j is the location of the traffic node, X i+1,j is the next traffic node position horizontally, X i,j+1 is the location of the longitudinal traffic node; Step S403: Based on the impedance model data, combined with steps S10 and S20, the state of charge of the electric vehicle when it arrives at the charging station is obtained. in is the state of charge of the electric vehicle when it arrives at the charging station, and P is the work done per unit time by the electric vehicle while driving on the road; The time model for reaching the charging station is: in is the time to reach the charging station, The time when charging demand occurs; Step S404: According to step S10, step S20, and step S30, the electric vehicle charging model is obtained as follows: in Charging time of large vehicles with fast charge, SOC max Charging maximum state of charge, Charging time for large vehicles with slow charging, Charging time for small vehicles with fast charging, Charging time for small vehicles with slow charging, P fast is the fast charging power, P slow is the slow charge power; Step S405, combining steps S10, S20, and S30 to obtain a charging load prediction model: P(T) Total =P(T) big +P(T) small Where P(T) big Charging power for large vehicles, P(T) small Charging power for small cars; When slow charging is selected, it is 1, and when not selected, it is 0; When fast charging is selected, it is 1, and when not selected, it is 0; P(T) Total is the total charging power at time T; The number of large vehicles that charge at charging stations, The number of small cars that charge at charging stations; Step S406, charging station location model; in is the initial position of the electric vehicle in the road network, X i,j is the location of the traffic node, X i+1,j is the next traffic node position horizontally, X i,j+1 is the location of the longitudinal traffic node; Step S407, combining step S401, step S402, step S403, step S404, step S405, and step S406, obtains a charging load prediction curve of the electric vehicle within one day, with a time interval of Δt, and simultaneously obtains a charging load prediction curve of each charging station within one day in space.

Images