CN110610260B - Driving energy consumption prediction system, method, storage medium and equipment - Google Patents

Abstract

The invention discloses a driving energy consumption prediction system, method, storage medium and equipment, wherein the prediction method includes acquiring historical working condition data of a planned driving route; constructing a training sample data set based on the historical working condition data; performing data processing on the training sample data set Training, establish the vehicle speed characteristic BP neural network model and the driving energy consumption BP neural network model; obtain the real-time working condition information on the planned driving route, input it into the vehicle speed characteristic BP neural network model for prediction, and obtain the future driving speed characteristic data, and then The vehicle speed characteristic data is input into the BP neural network model of driving energy consumption for prediction, and the future driving energy consumption data is obtained to realize the online prediction of driving energy consumption. The invention can realize online effective prediction of driving energy consumption under driving conditions of different road environments and traffic states, and helps to improve the efficiency of vehicle intelligent energy management.

Description

Driving energy consumption prediction system, method, storage medium and device

technical field

The present invention relates to the technical field of vehicle-mounted intelligent energy management in an intelligent transportation system and an intelligent network environment, and in particular to a system, method, storage medium and device for predicting energy consumption of vehicles on any planned route for intelligent vehicles.

Background technique

IEMS (Intelligent Energy Management System, Intelligent Energy Management System) is an inevitable requirement for the development of intelligent connected vehicles and ITS (Intelligent Transport System, Intelligent Transportation System), and its goal is to enable vehicles to adapt online in different driving scenarios Realize the efficient energy saving and optimal utilization of on-board energy, especially for the current new energy vehicles such as electric vehicles and hybrid vehicles. The vehicle energy consumption is mainly affected by the driving conditions in a certain road and traffic environment, so the key to the intelligent energy management system is to be able to achieve adaptive control of different driving conditions.

Since the traditional vehicle energy management system cannot know the driving conditions of the future vehicle, it is mainly based on the transient power point optimization control of the real-time operating point of the power system. However, the current intelligent energy management system can use the intelligent learning algorithm to realize the vehicle's driving speed prediction, driving power demand prediction or identification of driving conditions in a short period of time (usually within 3-5 minutes). Based on this, it can be further realized Adaptive operating condition control of vehicle energy, but these are only local optimal controls in the prediction time domain. Therefore, the current management of vehicle energy is mainly focused on instantaneous optimization and local optimization control, and there is little global optimization control of vehicle energy. Based on the prediction of driving energy consumption on the planned driving route in the future, the overall planning and optimal control of vehicle energy on the planned driving route can be further realized, and the utilization efficiency of vehicle energy can be greatly improved.

Contents of the invention

The present invention aims at the deficiencies of the above-mentioned prior art, based on the historical/real-time big data of road environment, traffic state and vehicle operation, online prediction is made on the driving energy consumption under different driving conditions on any planned driving route, so as to help realize the energy consumption of the vehicle. Global planning and optimization control.

One aspect of the present invention provides a driving energy consumption prediction system, including a data acquisition subsystem, an offline training subsystem and an online prediction subsystem; Parameters and vehicle operation data are collected and recorded; the offline training subsystem is used to divide the planned driving route into multiple road sections, extract and calculate road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of each road section , set up a sample data set; set up a BP neural network model, respectively train and verify the data set through the BP neural network, and obtain the verified BP neural network; Divide, extract the road environment parameters, traffic state parameters and mileage values of each road section, and predict the driving energy consumption on the planned driving route through the verified BP neural network.

Further, the offline training subsystem includes an offline data processing module and a model training module; wherein the offline data processing module is used to divide the planned driving route into multiple road sections, extract and calculate the road environment parameters and traffic conditions of the road sections Parameters, vehicle speed characteristic parameters and energy consumption value, establish the vehicle speed characteristic prediction sample data set and the driving energy consumption prediction sample data set respectively; The model training module is used to divide the sample data set into a training data set and a test data set, respectively A BP neural network model of vehicle speed characteristics and a BP neural network model of driving energy consumption are built, and the training data sets are respectively trained; the effectiveness of the trained BP neural network is verified through the test data set.

Further, the online prediction subsystem includes an online data processing module and a real-time prediction module; wherein the online data processing module is used to dynamically divide the planned driving route into real-time road sections, and extract road environment parameters and traffic state parameters of each road section and the road section mileage value; the real-time prediction module is used to predict the road section vehicle speed characteristic parameters on the future driving route in real time according to the parameters extracted by the vehicle speed characteristic BP neural network model obtained after training and the data processing module, Then use the predicted vehicle speed characteristic parameters as the input value of driving energy consumption BP driving energy consumption to predict the energy consumption value of each road section in real time in the future, and sum the energy consumption values of each road section to realize the driving energy consumption of the future driving route Prediction.

Another aspect of the present invention provides a driving energy consumption prediction method, including: acquiring historical working condition data of the planned driving route, including road environment parameters, traffic state parameters and vehicle operating data; constructing training sample data based on the historical working condition data set; carry out data training on the training sample data set, set up the vehicle speed characteristic BP neural network model and the driving energy consumption BP neural network model; obtain the real-time working condition information on the planned driving route, and input it to the vehicle speed characteristic BP neural network model Prediction in the vehicle to obtain the characteristic data of vehicle speed in the future, and then input the characteristic data of vehicle speed into the BP neural network model of driving energy consumption for prediction, obtain the data of future driving energy consumption, and realize the online prediction of driving energy consumption.

Further, the construction of the training sample data set specifically includes: dividing the planned driving route into multiple road sections, and extracting the characteristic parameters of the road section in units of a single road section, including road environment parameters, traffic state parameters, vehicle speed characteristic parameters, road section Traveling mileage value and vehicle driving energy consumption value; using stepwise linear regression method to analyze the influence relationship between the energy consumption value and the vehicle speed characteristic parameter, and determine the main vehicle speed characteristic parameter for the establishment of the prediction model.

Further, the division of the planned driving route into multiple road sections is specifically dividing the planned driving route into different traffic congestion levels, and a section of driving mileage with the same traffic congestion level is divided into a road section sample.

Further, the road environment parameters include road type, road gradient and road speed limit; the traffic state parameter is traffic congestion level; the mileage value of the road section is between the starting point of the planned driving route and the midpoint of the road section The distance of the vehicle speed characteristic parameter is the statistic of the vehicle speed sequence in a certain period of time, including average speed, average acceleration, speed standard deviation, average acceleration, acceleration standard deviation, acceleration time ratio, deceleration time ratio, uniform speed time ratio and Idle/stop time ratio, maximum vehicle speed and maximum acceleration.

Further, in the road environment parameters of the extracted road sections, the extraction method of road type and road speed limit parameter data is specifically: when the road type or road speed limit in the road sections divided according to the road section division method is unique, then the road section's The road type or road speed limit parameter data is the parameter value corresponding to the road type or road speed limit; when the road type or road speed limit in the road segment divided according to the road segment division method is not The speed parameter value is determined as follows: multiply the length ratio of different road types or road speed limits in the road section by the parameter value corresponding to the respective road type or road speed limit, and then sum all the parameter values to obtain the road section The final road type or road speed limit parameter value.

The present invention also provides a storage medium, including a program stored in the storage medium, and when the program is running, the device where the storage medium is located is controlled to execute the driving energy consumption prediction method described in any one of the above technical solutions.

The present invention also provides a driving energy consumption detection device, which includes a processor, the processor is used to run a program, and when the program is running, the method for predicting the driving energy consumption in any one of the above technical solutions is executed.

The present invention is based on real-time obtainable road environment, traffic status and driving mileage information to predict the driving energy consumption on the future planned driving route, which has strong adaptability and practicability to working conditions; based on the stepwise linear regression method, the main The vehicle speed characteristic parameters are screened to ensure that there is no collinearity between the selected vehicle speed characteristic parameters, which reduces the unnecessary input of the prediction model, and improves the prediction efficiency while ensuring the prediction accuracy; Driving energy consumption prediction can further help realize the overall planning and optimization of on-board energy, and greatly improve the utilization efficiency of on-board energy.

Description of drawings

The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention, and the schematic embodiments of the present invention and their descriptions are used to explain the present invention, but not to limit the present invention. In the attached picture,

Fig. 1 is a structural diagram of a vehicle energy consumption prediction system according to an embodiment of the present invention;

FIG. 2 is a flow chart of a method for predicting energy consumption during driving according to another embodiment of the present invention;

Fig. 3 is the histogram of the influence factor distribution of vehicle speed characteristics on energy consumption in the embodiment of Fig. 2;

Fig. 4 is the vehicle speed characteristic parameter prediction result figure in Fig. 2 embodiment;

Fig. 5 is a diagram of the prediction results of driving energy consumption based on the characteristic parameters of the real vehicle speed and the characteristic parameters of the predicted vehicle speed.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Example 1

This embodiment provides an intelligent prediction system for driving energy consumption. Figure 1 is a schematic diagram of an intelligent prediction system for driving energy consumption in this embodiment. Referring to Figure 1, the system includes a data acquisition subsystem, an offline training subsystem and an online prediction system. subsystem.

The data acquisition subsystem is used to collect and record road environment parameter data, traffic state parameter data and vehicle operation data on the planned driving route;

The offline training subsystem is used to divide the planned driving route into multiple road sections, extract and calculate the road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of the road sections, and analyze the vehicle speed characteristic parameters , establish a sample data set; build a BP neural network model, train and verify the data set through the BP neural network, and obtain a verified BP neural network.

The online prediction subsystem is used to divide the planned driving route into road sections, extract the roads of each road section, traffic characteristic parameters and road section mileage values, and analyze the traffic on the planned driving route by the BP neural network after the verification. Energy consumption is predicted.

Further, the data acquisition subsystem includes a road and traffic data acquisition module and a vehicle operation data acquisition module;

Among them, the road and traffic data acquisition module is used to collect and record the road environment parameters, traffic state parameters and mileage information of the planned driving route by using the vehicle GPS positioning device and the GIS information receiving device; the vehicle operation data acquisition module is used to It uses CAN bus and vehicle speed sensor to collect and record the running speed of the vehicle on the planned driving route.

Further, the offline training subsystem includes an offline data processing module and a model training module;

Among them, the offline data processing module is used to divide the planned driving route into multiple road sections, extract and calculate the road environment characteristic parameters, traffic state characteristic parameters, road mileage values, vehicle speed characteristic parameters and energy consumption values of the road sections. The regression method analyzes the influence relationship between the vehicle speed characteristic parameters and energy consumption, and respectively establishes a vehicle speed characteristic prediction sample data set and a driving energy consumption prediction sample data set;

The model training module is used to divide the sample data set into a training data set and a test data set, build a BP neural network model of vehicle speed characteristics and a BP neural network model of driving energy consumption, and train the training data set through the BP neural network; pass the test The dataset verifies the validity of the trained BP neural network.

Further, the online prediction subsystem includes an online data processing module and a real-time prediction module;

Among them, the online data processing module is used to dynamically divide the planned driving route into road sections, and extract the road environment parameters, traffic state parameters and road mileage values of each road section;

The real-time prediction module is used to predict the road section vehicle speed characteristic parameters on the future driving route in real time according to the vehicle speed characteristic BP neural network model obtained after training and the parameters extracted by the data processing module, and then use the predicted vehicle speed characteristic parameters as The input of driving energy consumption BP driving energy consumption predicts the energy consumption value of each road section in real time in the future, and finally sums the energy consumption values of each road section to complete the prediction of the driving energy consumption of the future driving route.

Example 2

As shown in Figure 2, a method for predicting energy consumption in driving, including:

Step 1. Obtain the historical working condition data of the planned driving route, including road environment parameters, traffic state parameters and vehicle operation data;

Use vehicle-mounted GPS positioning devices and GIS information receiving devices to collect road environment parameters such as road types, road slopes, road speed limits, and traffic status parameters such as traffic congestion levels; use CAN bus and speed sensors to collect vehicle operating data such as driving distance and speed wait.

Step 2. Construct a training sample data set based on the acquired raw data, specifically including the following steps:

Since the traffic status of different road sections on the planned driving route is different and time-varying, and the energy consumption of the vehicle operation is different under different traffic conditions, the driving mileage of a segment with the same traffic congestion level on the planned driving route is divided into is a road segment sample;

Further, for the divided road section samples, extract and calculate the road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of the road section, where

The mileage value of a road section is the distance between the midpoint of the road section and the starting point of the planned driving route; while the road type, road slope, road speed limit and traffic congestion level, these parameter values are directly obtained by the vehicle GPS positioning device and GIS information receiving device Generally, no further processing is required. However, the road type, road speed limit and traffic congestion level parameter values are discrete state values of limited types. After dividing the road sections according to the road section division method, the traffic congestion level parameter values of each road section can be uniquely determined, but there may be some If the road type or road speed limit of the road segment is not unique, at this time, the road type and road speed limit parameter values of the road segment need to be further processed as follows:

Multiply the length proportions of different road types or road speed limits in the road section by the parameter values corresponding to the respective road types or road speed limits, and then sum the above-mentioned parameter values multiplied by the length ratio to obtain the final road section road type or road speed limit parameter value.

The vehicle speed characteristic parameters are calculated from the vehicle speed sequence data of the road section samples. In this embodiment, in order to subsequently analyze the relationship between the vehicle speed characteristic parameters and energy consumption, 9 basic vehicle speed characteristic parameters are selected for calculation:

Among them, the characteristic parameters of vehicle speed specifically include: P ₀ is the ratio of idle speed/stopping time; P _a is the ratio of acceleration time; P _d is the ratio of deceleration time; P _y is the ratio of constant speed time; average speed V _m , speed standard deviation V _s , average Acceleration a _m , average deceleration d _m , acceleration standard deviation A _s .

First, assuming that the running time of the road section sample is T, the acceleration at each moment is calculated, and the idling/stopping time T ₀ , acceleration time T _a , deceleration time T _d and constant speed time _{Ty are} calculated:

In the formula, a _{i, i+1} is the acceleration of the i-th second and i+1 second, and the unit is m/s2; u _{i, i+1} is the speed of the i+1 second, and u _i is the speed of the i-th second , the unit is km/h; k is the number of all speed data points of the road section sample;

T ₀ = the total number of data points whose speed is 0 in the road section sample;

T _a = the total number of points whose acceleration is not less than 0.15m/s2 in the road section sample;

T _d = the total number of points whose acceleration is not greater than -0.15m/s2 in the road section sample;

T _y =TT ₀ -T _a -T _d

Further, all the characteristic parameters of vehicle speed are calculated as follows:

The energy consumption value is also calculated from the vehicle speed sequence data of the road section samples. First, the vehicle power balance equation is used to calculate the required power of the vehicle, and then the energy consumption formula is used to calculate the energy consumption within a period of time. The specific power and energy calculation formulas as follows:

Among them, m is the mass of the car, g is the acceleration of gravity, η _T is the transmission efficiency, i is the road gradient, _CD is the air resistance coefficient, A is the frontal area of the car, f is the rolling resistance coefficient, and δ is the rotation mass conversion coefficient, is the straight-line acceleration, P _e is the required power of the vehicle, u _a is the vehicle speed, and E is the energy.

Usually, in order to reduce the complexity of the prediction model and improve the prediction efficiency, the stepwise linear regression method is used to analyze the influence relationship between the various vehicle speed characteristic parameters and energy consumption, and the vehicle speed characteristic values that have a major impact on energy consumption are screened out to use Build up the predictive model with follow-up.

The basic principle of the above-mentioned stepwise regression is to introduce the variables into the model one by one. After each explanatory variable is introduced, the F-test must be performed, and the t-test must be performed on the explanatory variables that have been selected one by one. When it becomes no longer significant, it is deleted to ensure that only significant variables are included in the regression equation each time a new variable is introduced. This is an iterative process until neither significant explanatory variables are selected into the regression equation nor No insignificant explanatory variables were removed from the regression equation to ensure that the explanatory variables remaining in the model were both important and free from severe multicollinearity. The specific steps are as follows: First, use the explained variable (energy) to perform a simple regression on each considered explanatory variable (vehicle speed characteristic parameter), and then based on the regression equation corresponding to the explanatory variable that contributes the most to the explained variable, and then step by step Introduce the rest of the explanatory variables. The principle of the order of introduction is that this variable has a greater test value than other variables entering the model. A regression equation will be obtained at each step until the optimal regression equation is obtained, that is, the optimal set of explanatory variables is selected. .

The specific analysis results are shown in Table 1:

Table 1 Results of stepwise regression analysis

According to the analysis results in Table 1, the influence factors of each vehicle speed characteristic parameter on energy are calculated:

Among them, ΔR ² is the increase of the coefficient of determination R2 of the regression equation after introducing new variables at each step; R ² _l is the coefficient of determination of the optimal regression equation finally obtained by regression analysis.

As shown in Figure 4, it can be seen that the regression analysis finally retains the acceleration time ratio, average speed, speed standard deviation, average acceleration, average deceleration, and acceleration standard deviation; among them, the larger the impact factor value, the greater the impact on energy consumption. High, according to the distribution of influencing factors in Figure 4, select the first three speed characteristics with major influences for subsequent modeling, namely the acceleration time ratio, average speed and speed standard deviation.

After determining the vehicle speed characteristic parameters, the establishment of the final vehicle speed characteristic prediction sample data set and the driving energy consumption prediction sample data set is completed.

Step 3. Carry out data training on the sample data set to establish a vehicle speed characteristic BP neural network model and a driving energy consumption BP neural network model;

The vehicle speed characteristic prediction sample data set is trained by BP neural network, and the vehicle speed characteristic BP neural network model is established; the driving energy consumption prediction sample data set is trained by BP neural network, and the driving energy consumption BP neural network model is established.

Wherein, the steps of establishing the BP neural network include:

(1) Select the input variables of the network and determine the number of nodes in the input layer m;

(2) Determine the number of hidden layers and the number of hidden layer nodes;

(3) Determine the learning rate, initial weight, and initial threshold;

(4) Determine the number of nodes in the output layer;

(5) Training the neural network.

In this embodiment, for the vehicle speed characteristic BP neural network model, the input variables are road type, road speed limit, road gradient, traffic congestion level and mileage, the number of input layer nodes is 5, the number of hidden layers is 2, and the learning The ratio is 0.02, and the initial weight and initial threshold are both default values. Since the three vehicle speed characteristic parameters are predicted separately, the number of nodes in the output layer is 1.

For the BP neural network model of driving energy consumption, the input variables are acceleration time ratio, average speed and speed standard deviation, the number of nodes in the input layer is 3, the number of hidden layers is 2, the learning rate is 0.02, and the initial weight and initial threshold are both is the default value, the output variable is energy, and the number of nodes in the output layer is 1.

Among them, the number of nodes m in the hidden layer has three estimation methods as follows:

(1)

(2) m=log ₂ n

(3)

Among them, n is the number of nodes in the input layer, l is the number of nodes in the output layer, and δ is a constant between 0 and 10. The number of nodes in the hidden layer m is obtained by estimation method and trial and error method.

For the vehicle speed characteristic BP neural network model, the number of hidden layer nodes m is determined to be 25, and for the driving energy consumption BP neural network model, the number of hidden layer nodes m is determined to be 20.

The process of training the neural network is as follows:

For the BP neural network model of vehicle speed characteristics, 75% are randomly selected from the sample data set as training samples, 25% are used as test samples, and the road type, road speed limit, road gradient, traffic congestion level and driving mileage of the training samples are used as network input; Taking the vehicle speed characteristics of the training samples as the network output, using the standard BP model, the number of hidden layers is selected as 2, the number of nodes in the input layer is 5, the number of nodes in the output layer is 1, the number of nodes in the hidden layer is 25, and the transfer function of the first layer The tansig function is chosen, the transfer function of the second layer is the purelin function, the training function is the improved training function traindm with momentum gradient descent, and the training of the neural network is completed through data learning.

For the BP neural network model of driving energy consumption, randomly select 75% from the sample data set as training samples, 25% as test samples, and take the acceleration time ratio, average speed and speed standard deviation of the training samples as the network input; use the energy of the training samples Consumption is used as the network output, using the standard BP model, the number of hidden layers is selected as 2, the number of nodes in the input layer is 3, the number of nodes in the output layer is 1, the number of nodes in the hidden layer is 20, and the transfer function of the first layer is selected as the tansig function. The transfer function of the second layer is a purelin function, and the training function is an improved training function traindm with momentum gradient descent, and the training of the neural network is completed through data learning.

Next, use the test sample data to test the trained vehicle speed characteristic BP neural network model and driving energy consumption BP neural network model. The test results of the vehicle speed characteristic BP neural network model are shown in Figure 4 (a) and Figure 4 ( b) As shown in (c) in Figure 4, the test results of the BP neural network model of driving energy consumption are shown in (a) in Figure 5.

Using the predicted vehicle speed characteristic parameter data as the input of the BP neural network model of driving energy consumption, the driving energy consumption is predicted. The result is shown in (b) in Figure 5, and the vertical axis on the left in Figure 5 (b) is the road section The cumulative energy consumption, the vertical axis on the right is the absolute error of the cumulative energy consumption.

Among them, (a) in Figure 5 is the prediction result based on the real vehicle speed characteristic parameters as the input of the BP neural network model of driving energy consumption, and its final absolute error is about 100KJ, and (b) in Figure 5 is the predicted result The vehicle speed characteristic parameter is used as the prediction result of the input of the BP neural network model of driving energy consumption, and its final absolute error is about 160KJ. It can be seen that although the prediction accuracy of driving energy consumption based on the characteristic parameters of predicted vehicle speed is lower than the former, its prediction accuracy is still quite high, and the final relative error is about 7%.

Step 4. Obtain the real-time road and traffic information parameters on the planned driving route, input them into the vehicle speed characteristic BP neural network model to predict the future driving speed characteristic data, and then input the predicted vehicle speed characteristic data into the driving energy consumption BP neural network The model is used to predict the future driving energy consumption data and realize the online prediction of driving energy consumption.

Since the traffic status of different road sections on the planned driving route is different and time-varying, and the energy consumption of the vehicle operation is different under different traffic conditions, the driving mileage of a segment with the same traffic congestion level on the planned driving route is divided into for a road segment.

Based on the real-time traffic congestion level data on the planned route collected by the vehicle-mounted GPS positioning device and the GIS information receiving device, the planned driving route is dynamically divided into multiple road sections according to the method described, and other parameter data of each road section are further determined, including road types , road slope, road speed limit and road segment mileage value.

Among them, the mileage value of the road section is the distance between the midpoint of the road section and the starting point of the planned driving route; and the road type, road slope, road speed limit and traffic congestion level, these parameter values are the car GPS positioning device and GIS information receiving device Those obtained directly generally do not require further processing. However, the road type, road speed limit and traffic congestion level parameter values are discrete state values of limited types. After dividing the road sections according to the road section division method, the traffic congestion level parameter values of each road section can be uniquely determined, but there may be some If the road type or road speed limit of the road segment is not unique, at this time, the road type and road speed limit parameter value of the road segment need to be further processed as follows: the length ratio of different road types or road speed limits in the road segment Multiply by the parameter values corresponding to the respective road types or road speed limits, and then sum the parameter values multiplied by the length ratio to obtain the final road type or road speed limit parameter values for this road segment.

These obtained parameter data are input into the vehicle speed characteristic BP neural network model for prediction, and the vehicle speed characteristic data of each road section in the future are obtained. Then input the predicted vehicle speed characteristic data into the BP neural network model of driving energy consumption for prediction, and obtain the energy consumption data of each road section in the future. The total energy on the planned driving route can be obtained by summing the predicted energy consumption data of all road sections The energy consumption within any mileage on the planned driving route can be obtained by summing the predicted energy consumption data of any number of continuous road sections.

This embodiment also provides a storage medium, including a program stored in the storage medium, and when the program is running, the device where the storage medium is located is controlled to execute the driving energy consumption prediction method described in any one of the above technical solutions .

This embodiment also provides a driving energy consumption detection device, including a processor, the processor is used to run a program, and when the program is running, executes the driving energy consumption prediction method described in any one of the above technical solutions.

The technical means disclosed in the solutions of the present invention are not limited to the technical means disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features.

Claims (10)

1. A driving energy consumption prediction system is characterized in that: comprising a data acquisition subsystem, an offline training subsystem and an online prediction subsystem; wherein The data collection subsystem is used to collect and record road environment parameters, traffic state parameters and vehicle operation data on the planned driving route; The offline training subsystem is used to divide the planned driving route into multiple road sections, extract and calculate road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of each road section, and set up sample data sets; build BP neural network Model, through BP neural network, described data set is trained and verified respectively, obtains the BP neural network after verification; Described establishment sample data set comprises the following steps: S11. Divide a section of driving mileage with the same traffic congestion level on the planned driving route into a road section sample; S12. For the divided road section samples, extract and calculate the road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of the road section, where the mileage value of the road section is the distance between the midpoint of the road section and the starting point of the planned driving route , the road type, road slope, road speed limit and traffic congestion level are directly acquired by the vehicle GPS positioning device and GIS information receiving device; S13. Further process the road types and road speed limit parameter values of road sections: multiply the length ratios of different road types or road speed limits in this road section by the parameter values corresponding to the respective road types or road speed limits, and then divide each The sum of the parameter values multiplied by the length ratio is summed to obtain the final road type or road speed limit parameter value of the road section; the vehicle speed characteristic parameters specifically include: P ₀ is the ratio of idle speed/stopping time; P _a is the ratio of acceleration time; P _d is the deceleration time ratio; P _y is the constant speed time ratio; average speed V _m , speed standard deviation V _s , average acceleration a _m , average deceleration d _m , acceleration standard deviation A _s ; S14. Calculate the energy consumption value, first calculate the required power of the vehicle, and then calculate the energy consumption within a period of time; The online prediction subsystem is used to divide the planned driving route into road sections, extract the road environment parameters, traffic state parameters and road section mileage values of each road section, and analyze the traffic on the planned driving route through the BP neural network after the verification. Energy consumption is predicted. 2. The driving energy consumption prediction system according to claim 1, wherein the offline training subsystem includes an offline data processing module and a model training module; wherein The offline data processing module is used to divide the planned driving route into multiple road sections, extract and calculate the road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of the road sections, and respectively establish vehicle speed characteristic prediction sample data sets and driving Energy consumption prediction sample data set; The model training module is used to divide the sample data set into a training data set and a test data set, respectively build a vehicle speed characteristic BP neural network model and a driving energy consumption BP neural network model, and train the training data set respectively; The effectiveness of the trained BP neural network is verified through the test data set. 3. The driving energy consumption prediction system according to claim 1, wherein the online prediction subsystem includes an online data processing module and a real-time prediction module; wherein The online data processing module is used to divide the planned driving route into dynamic road sections in real time, and extract road environment parameters, traffic state parameters and road section mileage values of each road section; The real-time prediction module is used to predict the road section vehicle speed characteristic parameters on the future driving route in real time according to the vehicle speed characteristic BP neural network model obtained after training and the parameters extracted by the data processing module, and then use the predicted vehicle speed characteristic The parameter is used as the input value of the driving energy consumption BP driving energy consumption to predict the energy consumption value of each road section in the future in real time, and sum the energy consumption values of each road section to realize the prediction of the driving energy consumption of the future driving route. 4. A method for predicting energy consumption in driving, comprising the following steps: Obtain the historical working condition data of the planned driving route, including road environment parameters, traffic state parameters and vehicle operation data; Constructing a training sample data set based on the historical working condition data; establishing a sample data set includes the following steps: S11. Divide a section of driving mileage with the same traffic congestion level on the planned driving route into a road section sample; S12. For the divided road section samples, extract and calculate the road environment parameters, traffic state parameters, vehicle speed characteristic parameters and energy consumption values of the road section, where the mileage value of the road section is the distance between the midpoint of the road section and the starting point of the planned driving route , the road type, road slope, road speed limit and traffic congestion level are directly acquired by the vehicle GPS positioning device and GIS information receiving device; S13. Further process the road types and road speed limit parameter values of road sections: multiply the length ratios of different road types or road speed limits in this road section by the parameter values corresponding to the respective road types or road speed limits, and then divide each The sum of the parameter values multiplied by the length ratio is summed to obtain the final road type or road speed limit parameter value of the road section; the vehicle speed characteristic parameters specifically include: P ₀ is the ratio of idle speed/stopping time; P _a is the ratio of acceleration time; P _d is the deceleration time ratio; P _y is the constant speed time ratio; average speed V _m , speed standard deviation V _s , average acceleration a _m , average deceleration d _m , acceleration standard deviation A _s ; S14. Calculate the energy consumption value, first calculate the required power of the vehicle, and then calculate the energy consumption within a period of time; Carry out data training to described training sample dataset, set up vehicle speed feature BP neural network model and driving energy consumption BP neural network model; Obtain real-time working condition information on the planned driving route, input it into the vehicle speed characteristic BP neural network model for prediction, obtain the vehicle speed characteristic data of future driving, and then input the vehicle speed characteristic data into the driving energy consumption BP neural network model Carry out predictions, obtain future driving energy consumption data, and realize online prediction of driving energy consumption. 5. The driving energy consumption prediction method according to claim 4, wherein the construction of the training sample data set is specifically: dividing the planned driving route into a plurality of road sections, and extracting the characteristic parameters of the road section in units of a single road section , including road environment parameters, traffic state parameters, vehicle speed characteristic parameters, road section travel mileage value and vehicle travel energy consumption value; use stepwise linear regression method to analyze the influence relationship between the energy consumption value and the vehicle speed characteristic parameter, and determine The main vehicle speed characteristic parameters are used to establish the prediction model. 6. The driving energy consumption prediction method according to claim 5, wherein said dividing the planned driving route into a plurality of road sections is specifically dividing the planned driving route into different traffic congestion levels, and a section of driving with the same traffic congestion level The mileage is divided into a road segment sample. 7. The driving energy consumption prediction method according to claim 5, wherein said road environment parameters include road type, road gradient and road speed limit; said traffic state parameter is a traffic congestion level; said road section The mileage value is the distance between the starting point of the planned driving route and the midpoint of the road section; the vehicle speed characteristic parameter is the statistic of the vehicle speed sequence in a certain period of time, including average speed, average acceleration, speed standard deviation, average acceleration, Acceleration standard deviation, acceleration time ratio, deceleration time ratio, constant speed time ratio and idle/stop time ratio, maximum vehicle speed and maximum acceleration. 8. The driving energy consumption prediction method according to claim 6, characterized in that, in extracting the road environment parameters of road sections, the extraction method of road type and road speed limit parameter data is specifically: when dividing road sections according to the road section division method When the road type or road speed limit is unique, the road type or road speed limit parameter data of each road segment is the parameter value corresponding to the road type or road speed limit; when the road type or road speed limit in the road segment divided according to the road segment division method When the speed is not unique, the road type or road speed limit parameter value of this road segment is determined as follows: multiply the length ratio of different road types or road speed limits in this road segment by the corresponding road type or road speed limit parameter value, and then sum all the parameter values to obtain the final road type or road speed limit parameter value of the road segment. 9. A storage medium, characterized in that: it includes a program stored in the storage medium, and when the program is running, the device where the storage medium is located is controlled to perform the driving energy consumption described in any one of claims 4-8. method of prediction. 10. A driving energy consumption detection device, characterized in that it comprises a processor, the processor is used to run a program, and when the program is running, it executes the driving energy consumption prediction method according to any one of claims 4-8.

Images