Data prediction method and system

Abstract

The invention belongs to the technical field of data prediction, and particularly relates to a data prediction method and system. The neural network model of the first relation model is trained by acquiring a large number of environmental data of the electric vehicle in the parking state and the relation data of the corresponding battery SOC variation, and the training sample of the first relation model can be the corresponding environmental data of vehicles across the country, so that the method has the advantages of large sample space and strong generalization ability; and the motion data of the vehicle is trained through a neural network model of the second relation model, the prediction accuracy can be improved only by frequent training, and in the second relation model, the influence of the motion data on the SOC change is independently considered by stripping the influence of the environmental data on the SOC change, so that the prediction accuracy can be improved. Therefore, factors of environmental data can be omitted during training each time, and the calculation amount during model training is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of data prediction, and in particular, relates to a data prediction method and system. Background Art

[0002] Compared with traditional fuel vehicles, electric vehicles have advantages in driving economy and environmental friendliness, but there is still a large gap in power energy refueling time and convenience. Therefore, the "range anxiety" of mass consumers is an important factor that reduces the willingness to buy electric vehicles and confidence in using them. One of the reasons for "range anxiety" is that there is a large gap between the current electric vehicle's displayed range and the actual remaining range. The current industry's calculation of range is mostly based on the historical time domain, that is, based on the current battery remaining power and the average energy consumption in the past time domain, but this method has the risk of prediction not being consistent with reality.

[0003] With the development of deep learning technology, neural network algorithms have begun to be applied to the prediction method of electric vehicle range.

[0004] The Chinese patent with the authorization announcement number CN110091751B discloses a prediction method, device and medium for the range of electric vehicles based on deep learning, which collects historical basic data of the electric vehicles to be predicted and classifies the historical basic data; cleans the classified historical basic data to obtain historical cleaned data; builds a deep neural network model; uses the historical cleaned data as training data and uses a deep learning framework to train the deep neural network model; obtains the latest basic data of the electric vehicles to be predicted; inputs the basic data into the trained deep neural network model to obtain several range prediction results; and predicts the several range prediction results using a preset prediction formula to obtain the predicted range of the electric vehicles. This method uses a separate deep neural network model for each vehicle, which can maximize the accuracy of the range prediction of each vehicle.

[0005] In the above patent solution, in the neural network model, the amount of data required to be included in the data combination in the input layer is extremely large, specifically including a large amount of historical basic data combination, including electric vehicle motion data, battery data, motor data, motor controller data and environmental data; Motion data includes one or a combination of the following data: operating mode, vehicle speed, mileage, accelerator pedal travel, brake pedal travel, gear position, air conditioning status, charging status, vehicle status and braking force; battery data includes one or a combination of the following data: battery pack total voltage, battery pack total current, single cell voltage, single cell current, monitoring point temperature in the battery pack and SOC; motor data includes one or a combination of the following data: motor speed, motor torque and motor temperature; motor controller data includes one or a combination of the following data: motor controller input voltage, motor controller DC bus current and motor controller temperature; environmental data includes one or a combination of the following data: temperature outside the vehicle, longitude and latitude of the vehicle and altitude.

[0006] From the above description, we can see that there are many factors that affect the range of electric vehicles. If we need to improve the prediction accuracy of the range of electric vehicles, we need to comprehensively consider all the above factors that may affect the range. However, some of the above factors are data of common laws of this type of vehicle; Some of them are the characteristics of the driving habits of different drivers of the same model of vehicle. For example, the relationship between temperature, humidity and air pressure in environmental data and battery power consumption has nothing to do with the driving habits of the driver. The two are independent of each other. The influence of the driver's driving habits on the battery power consumption is very complex. Each driver's driving habits have their own characteristics. For example, when accelerating, whether to step on the accelerator pedal deeply or lightly, and when decelerating, the frequency and depth of the brake pedal will affect the battery power consumption. These subtle changes will affect the battery power consumption. These are all factors that affect the judgment of the cruising range. Even for a single driver, the driving habits at different times may be different. For example, there are differences in driving habits when in a hurry and when idle. And as the driver's driving experience increases, the driving habits will gradually change. On the other hand, the driver's driving path is also an important factor affecting power consumption and cruising range. That is, if the driver drives on a crowded road for a long time, the judgment of the cruising range is completely different from that of driving on a spacious road for a long time. However, even so, the driver's driving habits are also regular. After all, the change of habits requires a process, not an instantaneous change. Through the deep learning model, by continuously collecting the driver's driving data for training, the relationship between the instantaneous battery power consumption caused by driving habits can be obtained.

[0007] Therefore, it is urgent to improve the prediction method of electric vehicle range based on the collected data, and reduce the computational complexity of the training model while ensuring the accuracy of the prediction results. Summary of the invention

[0008] The technical problem to be solved by the present invention is: it is urgent to improve the prediction method of electric vehicle range based on the collected data, and reduce the calculation amount of the training model while ensuring the accuracy of the prediction result.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is: A data prediction method comprises the following steps: Acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; Inputting the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network based on a plurality of first sample data, wherein the first sample data includes the environmental data per unit time and the battery state of charge change value of the corresponding relationship; Obtaining motion data and environmental data per unit time and a battery state of charge change value △SOC1 of a single target vehicle in a driving state, wherein the motion data includes the vehicle's average speed, average acceleration, average motor torque, and average motor speed; Inputting the environmental data per unit time of the target vehicle into the first relationship model to obtain a battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; The motion data per unit time of the target vehicle is input into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; The environmental data and motion data of a plurality of unit times t of a discharge cycle of the target vehicle battery are obtained as test samples, and a plurality of △SOC2 and △SOC3 are obtained respectively through the first relationship model and the second relationship model, and the △SOC2 and △SOC3 obtained from each test sample are respectively added to obtain a plurality of △SOC 1’ , multiple △SOC are calculated 1’ The mean ; Get the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, get the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

[0010] Furthermore, in the above data prediction method, the second relationship model is established based on a generative adversarial network; wherein the generative adversarial network includes a generator and a classifier, and both the generator and the classifier include a convolutional neural network and a fully connected layer.

[0011] Furthermore, in the above data prediction method, the first relationship model is obtained by training a plurality of first sample data based on a neural network, and the training method of the first sample data includes the following steps: Obtaining a plurality of first sample data, wherein the first sample data includes corresponding environmental data per unit time and battery state of charge change values; For any first sample data, input the first sample data into the neural network, and output the matching probability corresponding to the first sample data; Calculating a loss value according to a preset loss function and a matching probability corresponding to the first sample data; If the loss value is less than a preset threshold, the training of the first relationship model is completed.

[0012] Furthermore, in the above data, the second relationship model is obtained by training a plurality of second sample data based on a neural network, and the training method of the second sample data includes the following steps: In the generative adversarial network structure, raw data is input into the generator; wherein the raw data is a subset of real data, and the real data includes motion data per unit time t and its corresponding battery state of charge change value △SOC3; The generator receives input data and outputs an output obtained by predicting the input data through a convolutional neural network output; The output of the generator is combined with the real data as the input of the classifier, so that the classifier can learn the real data and the generated data, and back propagation is used to improve the training accuracy; The output of the classifier is returned to train the generator.

[0013] The present invention also relates to a data prediction system, comprising: A first data acquisition module is used to acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; a first data processing module, configured to input the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network according to a plurality of first sample data, the first sample data including the environmental data per unit time and the battery state of charge change value of the corresponding relationship; The second data acquisition module is used to obtain the motion data and environmental data of a single target vehicle per unit time and the battery state of charge change value ΔSOC1 in the driving state, wherein the motion data includes the moving distance, speed average, acceleration average, motor torque average and motor speed average of the vehicle; The second data processing module is used to input the environmental data of the target vehicle per unit time into the first relationship model to obtain the battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; The third data processing module is used to input the motion data of the target vehicle per unit time into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; The fourth data processing module is used to obtain the environmental data and motion data of multiple unit times t of the target vehicle battery in the previous discharge cycle as test samples, obtain multiple △SOC2 and △SOC3 respectively through the first relationship model and the second relationship model, and add the △SOC2 and △SOC3 obtained from each test sample to obtain multiple △SOC 1’ , multiple △SOC are calculated 1’ The mean of The fifth data processing module is used to obtain the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, obtain the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

[0014] The beneficial effects of the present invention are as follows: in the data prediction method of the present invention, the relationship between environmental data and battery SOC changes does not change with the driving habits of the user, and the relationship data between the environmental data of a large number of electric vehicles in the parked state and the corresponding battery SOC changes are obtained to train the neural network model of the first relationship model. Since the training samples of the first relationship model can be the corresponding environmental data of vehicles across the country, it has the advantages of large sample space and strong generalization ability; and the relationship between the vehicle's motion data such as speed, acceleration, motor torque and speed and the battery SOC change varies due to the user's driving habits, and these data are collected separately by collecting the user's vehicle as training samples, and are trained by the neural network model of the second relationship model. Since the second sample data is personalized data, the generalization ability is weak, and continuous training is required to improve the generalization ability, and the sample space is limited, and frequent training is required to improve the accuracy of the prediction. In the second relationship model, by stripping off the influence of environmental data on SOC changes, the influence of motion data on SOC changes is considered separately, so the factor of environmental data can be omitted in each training, thereby reducing the amount of calculation when training the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of a flow chart of a data prediction method according to a specific embodiment of the present invention; Figure 2 It is a structural schematic diagram of a data prediction system according to a specific implementation mode of the present invention. DETAILED DESCRIPTION

[0016] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0017] Please refer to Figure 1 , Figure 1 The following is a flow chart of a data prediction method according to a specific embodiment of the present invention. The method comprises the following steps: S1: Acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; In the above steps, the ambient temperature, ambient humidity and ambient air pressure of the electric vehicle can be collected by the vehicle-mounted temperature sensor, vehicle-mounted humidity sensor and vehicle-mounted air pressure sensor, and can be collected according to the unit time. t collects multiple discrete data points and calculates the unit time through the processor The mean value of ambient temperature, ambient humidity and ambient air pressure within t; S2: Inputting the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network based on a plurality of first sample data, the first sample data including the environmental data per unit time and the battery state of charge change value of the corresponding relationship; In the above steps, the change value of the battery charge state per unit time can be obtained by obtaining the discrete data of the current point value of the lithium battery per unit time, calculating the average value of the current per unit time through the processor, multiplying it by the unit time, and dividing it by the rated capacity of the lithium battery, thereby calculating the change value of the battery charge state per unit time.

[0018] S3: Obtaining motion data and environmental data per unit time and battery state of charge change value ΔSOC1 of a single target vehicle in a driving state, wherein the motion data includes the vehicle's average speed, average acceleration, average motor torque, and average motor speed; In the above steps, speed and acceleration can be realized by any device equipped with a GPS system, such as a mobile device. Specifically, the mobile device obtains the driving trajectory data of the target vehicle based on the GPS system. For example, the GPS data of the target vehicle is obtained through the GPS system, which mainly includes the speed, position, time and other data of the target vehicle. Then, the acceleration corresponding to the target vehicle at the current moment can be calculated according to the speed, position and time, thereby obtaining the speed and acceleration corresponding to the target vehicle at the current moment. The torque value and motor speed of the motor can be obtained through the torque sensor and the speed sensor, which can be calculated according to the unit time. t collects multiple discrete data points and calculates the unit time through the processor The average ambient temperature, average ambient humidity, average ambient pressure, average speed, average acceleration, average motor torque and average motor speed within t.

[0019] S4: inputting the environmental data per unit time of the target vehicle into the first relationship model to obtain a battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; In the above steps, the change value of the battery charge state per unit time can be obtained by obtaining the discrete data of the current point value of the lithium battery per unit time, calculating the average value of the current per unit time through the processor, multiplying it by the unit time, and dividing it by the rated capacity of the lithium battery, thereby calculating the change value of the battery charge state per unit time.

[0020] S5: inputting the motion data per unit time of the target vehicle into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; Through the above calculation, the △SOC3 obtained is the SOC change value caused by the motion data of the user's vehicle after removing the influence of environmental factors on SOC. By using it as the second sample data, it is no longer necessary to consider the influence of environmental factors during training, thereby reducing the training calculation amount of the second relationship model.

[0021] S6: Acquire multiple environmental data and motion data of a unit time t of a discharge cycle of the target vehicle battery as test samples, obtain multiple △SOC2 and △SOC3 respectively through the first relationship model and the second relationship model, and add the △SOC2 and △SOC3 obtained from each test sample to obtain multiple △SOC 1’ , multiple △SOC are calculated 1’ The mean ; S7: Obtain the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, obtain the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

[0022] As a preferred implementation, the second relationship model is established based on a generative adversarial network; wherein the generative adversarial network includes a generator and a classifier, and both the generator and the classifier include a convolutional neural network and a fully connected layer.

[0023] In the above method, the generator and classifier in the generative adversarial network (GANs) are a combination of convolutional neural networks (CNNs) and fully connected layers. The generator uses convolutional neural networks to extract features, combines the features through fully connected layers, and outputs the predicted results through deconvolution layers. Its input is multiple unit times of driving status. The processed structure of the speed mean, acceleration mean, motor torque mean and motor speed mean within t is output as the corresponding SOC change value; the classifier uses a convolutional neural network to extract features and classifies through a fully connected layer to determine whether the input is real data or predicted data, and its output is a Boolean value. The structure of the generative adversarial network can improve the accuracy of energy consumption prediction.

[0024] As a preferred embodiment, the first relationship model is obtained by training a plurality of first sample data based on a neural network, and the training method of the first sample data includes the following steps: Obtaining a plurality of first sample data, wherein the first sample data includes corresponding environmental data per unit time and battery state of charge change values; For any first sample data, input the first sample data into the neural network, and output the matching probability corresponding to the first sample data; Calculating a loss value according to a preset loss function and a matching probability corresponding to the first sample data; If the loss value is less than a preset threshold, the training of the first relationship model is completed.

[0025] In the above steps, in order to prevent overfitting in the training process, L2 regularization and dropout algorithms are used to limit the model parameters of the convolutional neural network, increase the generalization ability of the convolutional neural network, and improve the accuracy. Among them, the preset loss function of the convolutional neural network adds an L2 regularization term. In order to avoid the gradient disappearance or gradient explosion caused by the deepening of the number of layers of the convolutional neural network and the slow convergence speed, the batch normalization (BatchNormalization, referred to as BN) algorithm can be used to standardize the input of certain layers or all layers, thereby fixing the mean and variance of the input signal of each layer, so that the input of each layer of the convolutional neural network maintains the same distribution during the training process, which simplifies the construction of deep neural networks and speeds up the learning convergence speed.

[0026] As a preferred embodiment, the second relationship model is obtained by training a plurality of second sample data based on a neural network, and the training method of the second sample data comprises the following steps: In the generative adversarial network structure, raw data is input into the generator; wherein the raw data is a subset of real data, and the real data includes motion data per unit time t and its corresponding battery state of charge change value △SOC3; The generator receives input data and outputs an output obtained by predicting the input data through a convolutional neural network output; The output of the generator is combined with the real data as the input of the classifier, so that the classifier can learn the real data and the generated data, and back propagation is used to improve the training accuracy; The output of the classifier is returned to train the generator.

[0027] In the second relational model, the generator and classifier are trained separately. The loss function of the classifier is the cross entropy of the classification result, while the loss function of the generator is [the sum of the squares of the error between the generator's output and the expected value] plus [the result of inputting the generator's output into the classifier]. The two are not added directly, but added according to certain weights. In this way, the purpose of using the results of the classifier to train the generator is achieved.

[0028] Please refer to Figure 2 The present invention also relates to a data prediction system, comprising: A first data acquisition module is used to acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; a first data processing module, configured to input the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network according to a plurality of first sample data, the first sample data including the environmental data per unit time and the battery state of charge change value of the corresponding relationship; The second data acquisition module is used to obtain the motion data and environmental data of a single target vehicle per unit time and the battery state of charge change value ΔSOC1 in the driving state, wherein the motion data includes the moving distance, speed average, acceleration average, motor torque average and motor speed average of the vehicle; The second data processing module is used to input the environmental data of the target vehicle per unit time into the first relationship model to obtain the battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; The third data processing module is used to input the motion data of the target vehicle per unit time into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; The fourth data processing module is used to obtain the environmental data and motion data of multiple unit times t of the target vehicle battery in the previous discharge cycle as test samples, obtain multiple △SOC2 and △SOC3 respectively through the first relationship model and the second relationship model, and add the △SOC2 and △SOC3 obtained from each test sample to obtain multiple △SOC 1’ , multiple △SOC are calculated 1’ The mean of The fifth data processing module is used to obtain the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, obtain the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

[0029] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A data prediction method, characterized in that: The following steps are involved: Acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; Inputting the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network based on a plurality of first sample data, wherein the first sample data includes the environmental data per unit time and the battery state of charge change value of the corresponding relationship; Obtaining motion data and environmental data per unit time and a battery state of charge change value △SOC1 of a single target vehicle in a driving state, wherein the motion data includes the vehicle's average speed, average acceleration, average motor torque, and average motor speed; Inputting the environmental data per unit time of the target vehicle into the first relationship model to obtain a battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; The motion data per unit time of the target vehicle is input into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; The environmental data and motion data of multiple unit times t of the last discharge cycle of the target vehicle battery are obtained as test samples, and multiple △SOC2 and △SOC3 are obtained respectively through the first relationship model and the second relationship model, and the △SOC2 and △SOC3 obtained from each test sample are added together to obtain multiple △SOC 1’ , multiple △SOC are calculated 1’ The mean ; Get the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, get the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

2. The data prediction method according to claim 1, characterized in that: The second relationship model is established based on a generative adversarial network; wherein the generative adversarial network includes a generator and a classifier, and both the generator and the classifier include a convolutional neural network and a fully connected layer.

3. The data prediction method according to claim 1, characterized in that: The first relationship model is obtained by training a plurality of first sample data based on a neural network, and the training method of the first sample data includes the following steps: Obtaining a plurality of first sample data, wherein the first sample data includes corresponding environmental data per unit time and battery state of charge change values; For any first sample data, input the first sample data into the neural network, and output the matching probability corresponding to the first sample data; Calculating a loss value according to a preset loss function and a matching probability corresponding to the first sample data; If the loss value is less than a preset threshold, the training of the first relationship model is completed.

4. The data prediction method according to claim 1, characterized in that: The second relationship model is obtained by training a plurality of second sample data based on a neural network, and the training method of the second sample data includes the following steps: In the generative adversarial network structure, raw data is input into the generator; wherein the raw data is a subset of real data, and the real data includes motion data per unit time t and its corresponding battery state of charge change value △SOC3; The generator receives input data and outputs an output obtained by predicting the input data through a convolutional neural network output; The output of the generator is combined with the real data as the input of the classifier, so that the classifier can learn the real data and the generated data, and back propagation is used to improve the training accuracy; The output of the classifier is returned to train the generator.

5. A data prediction system, characterized in that: include: A first data acquisition module is used to acquire environmental data per unit time and battery state of charge change values ​​of multiple vehicles in a parked state, wherein the vehicle environmental data includes environmental temperature, environmental humidity and environmental pressure; a first data processing module, configured to input the environmental data per unit time into a first relationship model to obtain a battery state of charge change value corresponding to the environmental data per unit time; the first relationship model is obtained by training a neural network according to a plurality of first sample data, the first sample data including the environmental data per unit time and the battery state of charge change value of the corresponding relationship; The second data acquisition module is used to obtain the motion data and environmental data of a single target vehicle per unit time and the battery state of charge change value ΔSOC1 in the driving state, wherein the motion data includes the moving distance, speed average, acceleration average, motor torque average and motor speed average of the vehicle; The second data processing module is used to input the environmental data per unit time of the target vehicle into the first relationship model to obtain a battery state of charge change value △SOC2 corresponding to the environmental data per unit time t; The third data processing module is used to input the motion data of the target vehicle per unit time into the second relationship model to obtain a battery state of charge change value △SOC3 corresponding to the motion data per unit time t, wherein △SOC3=△SOC1-△SOC2; the second relationship model is obtained based on a neural network trained according to a plurality of second sample data, wherein the second sample data includes the motion data per unit time t of the corresponding relationship and its corresponding battery state of charge change value △SOC3; The fourth data processing module is used to obtain the environmental data and motion data of multiple unit times t of the target vehicle battery in the previous discharge cycle as test samples, obtain multiple △SOC2 and △SOC3 respectively through the first relationship model and the second relationship model, and add the △SOC2 and △SOC3 obtained from each test sample to obtain multiple △SOC 1’ , multiple △SOC are calculated 1’ The mean of The fifth data processing module is used to obtain the mean value v of the historical driving speed of the target vehicle, calculate the average driving distance S of the target vehicle per unit time t through the formula S=vt, obtain the current battery state of charge value SOC of the target vehicle, and calculate the average driving distance S of the target vehicle per unit time t through the formula S obtains the predicted value of the cruising range.

Images