A regional energy consumption prediction method and system for new energy vehicles
By acquiring basic energy consumption data in the laboratory and combining it with road test data, and using RBF neural networks and energy consumption difference factors for iterative optimization, a regionalized coupled energy consumption prediction model was constructed. This solved the problems of accuracy and regional adaptability in energy consumption evaluation of new energy vehicles, and achieved high-precision energy consumption prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI MOTOR VEHICLE INSPECTION CERTIFICATION & TECH INNOVATION CENT CO LTD
- Filing Date
- 2026-02-06
- Publication Date
- 2026-06-09
AI Technical Summary
Existing technologies cannot fully reflect the energy consumption levels of new energy vehicles in complex and ever-changing real-world usage scenarios. Laboratory tests are disconnected from actual road conditions, road tests are costly and have poor repeatability, and there is a lack of scientific and objective regional energy consumption prediction methods.
By acquiring basic energy consumption data from a standard laboratory, a hub energy consumption prediction model is trained. This model is then discretized using road test data. An RBF neural network is used to predict standard laboratory energy consumption. An energy consumption difference factor is introduced for iterative optimization, and a regionalized coupled energy consumption prediction model is constructed. This model is then dynamically corrected by incorporating multi-dimensional regional feature parameters.
It achieves high accuracy and reliability in predicting the energy consumption of new energy vehicles, solves the problem of the disconnect between laboratory testing and actual road conditions, provides scientific and accurate regional energy consumption evaluation, and improves the pertinence and guidance of the evaluation.
Smart Images
Figure CN122171222A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of new energy vehicle technology, and in particular to a method and system for predicting the regional energy consumption of new energy vehicles. Background Technology
[0002] With the rapid global adoption of new energy vehicles, their energy consumption levels have become a key indicator for consumers' purchasing decisions, manufacturers' product optimization, and regulatory policy-making. However, the real-world usage scenarios of new energy vehicles are extremely complex, influenced by multiple factors such as geography, climate, traffic, and driving behavior, leading to significant challenges in energy consumption assessment. While current mainstream energy consumption assessment methods have some value, they have significant limitations and cannot fully reflect the vehicle's performance in real-world environments. The energy consumption levels in these complex and ever-changing real-world usage scenarios have become a focus of attention for consumers, manufacturers, and regulatory agencies. Current mainstream energy consumption assessment methods have the following limitations: 1) Laboratory standard cycle test: The advantage is that it is repeatable and highly comparable, but its fixed working conditions cannot fully reflect the ever-changing actual road conditions, such as frequent starts and stops, slope changes, extreme temperatures, air conditioning use, and individual driving behavior differences, resulting in a large gap between laboratory data and actual user energy consumption.
[0003] 2) Simple road testing: While it can reflect real-world conditions, it is costly, time-consuming, and has poor repeatability. Furthermore, it is heavily influenced by specific times, locations, and drivers, making it difficult to establish universally applicable evaluation standards. Although large datasets can be easily accumulated using tools like Gaode and Baidu Maps, the lack of standardized anchors makes it impossible to evaluate and compare products on a large scale, especially for constantly updated vehicle models.
[0004] 3) Insufficient existing regional energy consumption research: Current energy consumption evaluations mostly adopt a method of subjective weighting based on different operating temperatures and test scenarios to obtain a uniform nominal value. Objectively, this method does not fully consider the significant impact of geographical climate and traffic conditions in different regions on energy consumption, and lacks a scientific, objective, and digital method to predict the real energy consumption level of a certain vehicle model in a specific region.
[0005] For example, patent application CN118654901A discloses a virtual evaluation method and device for the actual road energy consumption of new energy vehicles. This method constructs an evaluation route, generates a database of actual road conditions including months, weather conditions, road conditions, and usage periods, uses vehicle models for simulation testing or real-vehicle testing to obtain data, and introduces the nominal energy consumption dispersion coefficient and average comprehensive energy consumption as evaluation indicators to determine the vehicle's energy consumption level. Although this method can cover multiple cities and scenarios to a certain extent, it can only obtain energy consumption statistics based on limited discrete operating condition sampling, and cannot meet the requirements of new energy vehicle energy consumption evaluation for accuracy, regionalization, and digital intelligence. Summary of the Invention
[0006] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a regional energy consumption prediction method and system for new energy vehicles, which significantly improves the accuracy and reliability of regional energy consumption assessment.
[0007] The objective of this invention can be achieved through the following technical solutions: A method for predicting the regional energy consumption of new energy vehicles includes the following steps: In a standard laboratory environment, acquire basic energy consumption data for the target new energy vehicle; A hub energy consumption prediction model is trained based on the energy consumption data of the target new energy vehicle. Collect road test data of the target new energy vehicle in typical road test areas with significant differences; The road test data is discretized to obtain labeled road test data; Based on the road test data processed by the labeling, the laboratory standard energy consumption of each typical road test area is predicted by the hub energy consumption prediction model. The energy consumption difference factor of the target new energy vehicle in each typical road test area is calculated based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory. The energy consumption difference factors of the target new energy vehicle in each typical road test area are correlated with the corresponding regional characteristic parameters, and the energy consumption difference factors of the target new energy vehicle in each typical road test area are iteratively optimized based on the correlation analysis results. Based on the iteratively optimized energy consumption difference factor, the hub energy consumption prediction model is corrected to construct a regionalized coupled energy consumption prediction model. The regional feature parameters of the test area are obtained from the preset regional feature parameter library and used as the input of the regionalized coupled energy consumption prediction model to predict the coupled predicted energy consumption of the target new energy vehicle in the test area.
[0008] Furthermore, the energy consumption baseline data includes ambient set temperature, standard cycle operating condition energy consumption, and various vehicle selection modes.
[0009] Furthermore, the road test data includes vehicle driving status, environmental data, and vehicle setting data. The vehicle driving status includes actual vehicle speed, energy consumption data, and vehicle CAN bus data. The environmental data includes temperature, humidity, slope, and altitude. The vehicle setting data includes vehicle mode, energy recovery mode, and air conditioning mode.
[0010] Furthermore, the specific steps for discretizing the road test data to obtain labeled road test data include: The energy window method is adopted, and the continuous time series of road test measurement data is divided into a series of continuous energy windows with a preset energy value as the energy window unit; For each energy window, the corresponding environmental parameters, vehicle setting parameters, and driving status parameters are labeled to obtain labeled road test data.
[0011] Furthermore, the hub energy consumption prediction model is trained using the RBF neural network method, and the laboratory standard energy consumption for each road test area is predicted by the hub energy consumption prediction model as follows: In the formula, For the target new energy vehicles in the first Laboratory standard energy consumption for each road test area To connect the first The weights of the radial basis functions to the output layer The total number of radial basis functions. For radial basis functions, In order to target the The labeled road test data vectors for each road test area. For the first The center vector of each radial basis function This is a bias term.
[0012] Furthermore, the energy consumption difference factor is: In the formula, For the target new energy vehicles in the first Energy consumption difference factors in a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. For the target new energy vehicles in the first Laboratory standard energy consumption for a typical road test area.
[0013] Furthermore, the regional characteristic parameters include climate and geographical parameters, traffic condition parameters, new energy vehicle user quantity parameters, and driving behavior parameters. The specific formula for iteratively optimizing the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results is as follows: In the formula, For the first The typical road test area in the first Optimize the energy consumption difference factor in round iteration. , and For dynamic weighting coefficients, For the first The typical road test area in the first The original energy consumption difference factor in the round of iteration, The moving average of the historical optimization energy consumption difference factor. For the first The typical road test area in the first The quantization correction term for the regional feature parameters in the round of iteration.
[0014] Furthermore, the loss function of the regionalized coupled energy consumption prediction model is: In the formula, The loss function for the regionalized coupled energy consumption prediction model is... This represents the total number of typical road test areas. To weight the mean square error between predicted energy consumption and measured energy consumption, For the regionalized coupled energy consumption prediction model in the first Coupled predicted energy consumption for a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. The KL divergence weights represent the differences in the distribution of regional characteristics. Let KL divergence be the KL divergence. The empirical distribution of characteristic parameters of the region to be tested. For the first The baseline distribution of characteristic parameters of a typical test area. The final optimized Euclidean distance weights between the energy consumption difference factor and the baseline energy consumption difference factor are used. For the first The final optimized energy consumption difference factor for a typical road test area For the first Benchmark energy consumption difference factor for a typical road test area The regularization intensity coefficient is . For regularization terms of model parameters, The standard benchmark energy consumption for target new energy vehicles For regional feature coupling function, For the first A set of regional characteristic parameters for a typical road test area. These are the model parameters.
[0015] Furthermore, the coupled predicted energy consumption is transformed into a regionalized energy consumption score through linear fitting, and the regionalized energy consumption score is: In the formula, For the target new energy vehicle in the test area Regionalized energy consumption rating The coefficient representing the generally accepted increase in energy consumption during road tests compared to laboratory standard energy consumption. The benchmark score is 100 points. The coupled prediction of energy consumption of the target new energy vehicle in the test area.
[0016] According to another aspect of the present invention, a regionalized energy consumption prediction system for new energy vehicles is provided, comprising: The energy consumption basic data acquisition module is used to acquire the energy consumption basic data of the target new energy vehicle in a standard laboratory environment. The hub energy consumption prediction model training module is used to train the hub energy consumption prediction model based on the energy consumption basic data of the target new energy vehicle. The road test data acquisition module is used to collect road test data of the target new energy vehicle in typical road test areas with significant differences. The data labeling and processing module is used to discretize the road test measured data to obtain labeled road test measured data. The laboratory standard energy consumption acquisition module is used to predict the laboratory standard energy consumption of each typical road test area based on the road test measured data processed by the labeling and through the hub energy consumption prediction model. The energy consumption difference factor acquisition module is used to calculate the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory. The energy consumption difference factor iterative optimization module is used to perform correlation analysis between the energy consumption difference factor of the target new energy vehicle in each typical road test area and the corresponding regional characteristic parameters, and to iteratively optimize the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results. The regionalized coupled energy consumption prediction model construction module is used to correct the hub energy consumption prediction model based on the iteratively optimized energy consumption difference factor and construct the regionalized coupled energy consumption prediction model. The coupled energy consumption prediction module is used to obtain the regional feature parameters of the area to be tested from a preset regional feature parameter library as the input of the regionalized coupled energy consumption prediction model, and to predict the coupled predicted energy consumption of the target new energy vehicle in the area to be tested.
[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention is based on a framework that couples laboratory standard data with real road data. By introducing energy consumption difference factors for data calibration and iterative optimization to quantify the deviation between theoretical and actual values, and by using a machine learning model to associate multi-dimensional regional feature parameters to dynamically correct predictions, it achieves high-precision prediction of regional energy consumption of new energy vehicles. This solves the problem that existing technologies suffer from large deviations between prediction results and real user experience due to laboratory testing being detached from actual road conditions and sparse road test samples, and significantly improves the accuracy and reliability of regional energy consumption evaluation.
[0018] 2. This invention integrates key variables affecting energy consumption by constructing a regional feature database containing multi-dimensional parameters such as geography, climate, and transportation. It generates coupled predicted energy consumption and energy consumption scores for the region under test through a regionalized coupled energy consumption prediction model, achieving accurate regional adaptation of energy consumption evaluation. This solves the problems of existing evaluation standards ignoring regional differences and having weak guidance for results, enabling the evaluation results to provide more targeted scientific basis for energy consumption evaluation in various regions. Attached Figure Description
[0019] Figure 1 This is a flowchart illustrating a regionalized energy consumption prediction method for new energy vehicles proposed in this invention. Figure 2 This is a schematic diagram of the structure of a regionalized energy consumption prediction system for new energy vehicles proposed in this invention.
[0020] Legend: 1. Basic energy consumption data acquisition module; 2. Hub energy consumption prediction model training module; 3. Road test measured data acquisition module; 4. Data labeling and processing module; 5. Laboratory standard energy consumption acquisition module; 6. Energy consumption difference factor acquisition module; 7. Energy consumption difference factor iterative optimization module; 8. Regionalized coupled energy consumption prediction model construction module; 9. Coupled energy consumption prediction module. Detailed Implementation
[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.
[0022] The following English abbreviations are involved: Radial Basis Function (RBF) Example 1 This embodiment provides a method for predicting the regionalized energy consumption of new energy vehicles, such as... Figure 1 As shown, it includes the following steps: S1. Obtain basic energy consumption data for the target new energy vehicle in a standard laboratory environment.
[0023] Acquire basic energy consumption data of the target new energy vehicle under various standard cycle conditions (such as CLTC-P and WLTC) on a drum test bench in a standard laboratory environment, including but not limited to ambient temperature setting, standard cycle energy consumption (vehicle speed-time curve, instantaneous energy consumption, average energy consumption, battery voltage and current), and vehicle selection mode.
[0024] S2. Train the hub energy consumption prediction model based on the energy consumption data of the target new energy vehicle.
[0025] The in-vehicle energy consumption prediction model is trained using the RBF neural network method based on the energy consumption baseline data of the target new energy vehicle. The number of nodes in the input layer equals the dimension of the input vector, i.e., the number of parameters in the energy consumption baseline data. The hidden layer uses radial basis functions as activation functions, and the output layer is a single node, outputting the predicted laboratory standard energy consumption. The training process includes: The centers of the radial basis functions are determined, and a clustering algorithm (such as K-Means) is used to cluster all training samples (i.e., multiple sets of basic energy consumption data). The center point of each cluster obtained after clustering serves as the center of a radial basis function. In this way, each radial basis function is responsible for responding to a local region of the input space.
[0026] The width of the radial basis functions is determined. Each radial basis function requires a width parameter (or standard deviation), which determines the range of sensitivity of the function to changes in the input. Typically, the width can be determined based on the distance between the center points to ensure smooth coverage.
[0027] To calculate the hidden layer output, for each input sample, calculate its Euclidean distance to each center, substitute this distance into the radial basis function, and obtain the output of that hidden node. The closer the input is to the center, the stronger the output of that node.
[0028] The output layer weights and biases are calculated, and the output of the hidden layers has a linear relationship with the final energy consumption value. The goal of this step is to find a set of weights and biases that make the network's predictions as close as possible to the actual measured energy consumption values in the laboratory.
[0029] S3. Collect road test data of the target new energy vehicle in typical road test areas with significant differences.
[0030] For the target new energy vehicle model, actual road tests are conducted in at least two significantly different typical regions (such as a cold region and a temperate region) according to the methods of urban areas, suburbs and highways. The road test data are collected, including vehicle driving status, environmental data and vehicle setting data. Vehicle driving status includes actual vehicle speed, energy consumption data and vehicle CAN bus data (such as motor torque, battery SOC, air conditioning power). Environmental data includes temperature, humidity, slope and altitude. Vehicle setting data includes vehicle mode (economy, comfort, sport, etc.), energy recovery mode (low, medium and high) and air conditioning mode (economy mode, air conditioning temperature, air outlet direction).
[0031] S4. Discretize the road test data to obtain labeled road test data.
[0032] The specific steps for discretizing the road test data to obtain labeled road test data include: The energy window method is adopted, and the continuous time series of road test measurement data is divided into a series of continuous energy windows with a preset energy value as the energy window unit; For each energy window, the corresponding environmental parameters, vehicle setting parameters, and driving status parameters are labeled to obtain labeled road test data.
[0033] S5. Based on the marked road test data, the laboratory standard energy consumption of each typical road test area is predicted by the hub energy consumption prediction model.
[0034] The laboratory standard energy consumption for each road test area was predicted using the hub energy consumption prediction model: In the formula, For the target new energy vehicles in the first Laboratory standard energy consumption for each road test area To connect the first The weights of the radial basis functions to the output layer The total number of radial basis functions. For radial basis functions, In order to target the The labeled road test data vectors for each road test area. For the first The center vector of each radial basis function This is a bias term.
[0035] S6. The energy consumption difference factor of the target new energy vehicle in each typical road test area is calculated based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory.
[0036] The energy consumption difference factor is: In the formula, For the target new energy vehicles in the first Energy consumption difference factors in a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. For the target new energy vehicles in the first Laboratory standard energy consumption for a typical road test area.
[0037] S7. Correlation analysis is performed between the energy consumption difference factor of the target new energy vehicle in each typical road test area and the corresponding regional characteristic parameters, and the energy consumption difference factor of the target new energy vehicle in each typical road test area is iteratively optimized based on the correlation analysis results.
[0038] Regional characteristic parameters include climate and geographical parameters, traffic condition parameters, number of new energy vehicle users, and driving behavior parameters.
[0039] Climate geography parameters include annual average temperature, temperature range distribution probability, average altitude, and slope distribution.
[0040] Traffic condition parameters include regional average vehicle speed, congestion index, and typical road network structure (ratio of urban expressways, arterial roads, and secondary arterial roads).
[0041] The number of new energy vehicle users is the number of new energy vehicles in each region.
[0042] The driving behavior parameters are the regional average frequency of rapid acceleration and deceleration obtained based on big data analysis.
[0043] The specific formula for iteratively optimizing the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results is as follows: In the formula, For the first The typical road test area in the first Optimize the energy consumption difference factor in round iteration. , and For dynamic weighting coefficients, For the first The typical road test area in the first The original energy consumption difference factor in the round of iteration, The moving average of the historical optimization energy consumption difference factor. For the first The typical road test area in the first The quantization correction term for the regional feature parameters in the round of iteration.
[0044] S8. Based on the iteratively optimized energy consumption difference factor, the hub energy consumption prediction model is corrected to construct a regionalized coupled energy consumption prediction model.
[0045] The loss function of the regionalized coupled energy consumption prediction model is: In the formula, The loss function for the regionalized coupled energy consumption prediction model is... This represents the total number of typical road test areas. To weight the mean square error between predicted energy consumption and measured energy consumption, For the regionalized coupled energy consumption prediction model in the first Coupled predicted energy consumption for a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. The KL divergence weights represent the differences in the distribution of regional characteristics. Let KL divergence be the KL divergence. The empirical distribution of characteristic parameters of the region to be tested. For the first The baseline distribution of characteristic parameters of a typical test area. The final optimized Euclidean distance weights between the energy consumption difference factor and the baseline energy consumption difference factor are used. For the first The final optimized energy consumption difference factor for a typical road test area For the first Benchmark energy consumption difference factor for a typical road test area The regularization intensity coefficient is . For regularization terms of model parameters, The standard benchmark energy consumption for target new energy vehicles For regional feature coupling function, For the first A set of regional characteristic parameters for a typical road test area. These are the model parameters. The process for obtaining the baseline energy consumption difference factor is as follows: By using clustering algorithms based on the regional feature parameters of all regions in a regional feature parameter database, all historical test areas are divided into several regional categories with similar characteristics. For example, they may be clustered into regional categories such as "hot and congested urban areas", "mild and highway suburbs", and "high-altitude and cold mountainous areas".
[0046] When it is necessary to [do something] When performing iterative optimization for the current round in a typical road test area, according to the... The regional characteristic parameters of a typical road test area are used to determine which regional category it belongs to in the above cluster analysis. Historical optimization difference factor data for all areas classified into that category are identified. Then, a moving average of these historical data is calculated as the baseline energy consumption difference factor.
[0047] S9. Obtain the regional feature parameters of the area to be tested from the preset regional feature parameter library as the input of the regionalized coupled energy consumption prediction model, and predict the coupled predicted energy consumption of the target new energy vehicle in the area to be tested.
[0048] The regional feature parameter library preset in this embodiment stores quantitative feature parameters of various regions around the world. This library is dynamically updated, and the data comes from publicly available geographic information systems (GIS), meteorological data, traffic big data from navigation platforms, and fleet management data, etc.
[0049] The model executes the final prediction formula as follows: In the formula, To predict the energy consumption of the target new energy vehicle in the test area through coupling, The standard benchmark energy consumption for target new energy vehicles For regional feature coupling function, The set of regional feature parameters for the region to be tested. For model parameters, This is the final optimized energy consumption difference factor for the region to be tested.
[0050] The coupled predicted energy consumption is transformed into a regionalized energy consumption score through linear fitting. The regionalized energy consumption score is as follows: In the formula, For the target new energy vehicle in the test area Regionalized energy consumption rating The coefficient representing the generally accepted increase in energy consumption during road tests compared to laboratory standard energy consumption. The benchmark score is 100 points. The coupled prediction of energy consumption of the target new energy vehicle in the test area.
[0051] This embodiment uses a weighted prediction comprehensive score based on the number of new energy vehicles in each region. Based on the above model, it sorts the energy consumption scores of multiple models in specific regions, generates a regional energy consumption score ranking, provides consumers with a more valuable car purchase guide, and can also generate a comprehensive score ranking to form a comprehensive and complex regional applicability evaluation ranking.
[0052] Example 2 This embodiment provides a regionalized energy consumption prediction system for new energy vehicles, such as... Figure 2 As shown, it includes: Energy consumption basic data acquisition module 1 is used to acquire the energy consumption basic data of the target new energy vehicle in a standard laboratory environment; Hub energy consumption prediction model training module 2 is used to train the hub energy consumption prediction model based on the energy consumption basic data of the target new energy vehicle. The road test data acquisition module 3 is used to collect road test data of the target new energy vehicle in typical road test areas with significant differences. Data labeling and processing module 4 is used to discretize the road test measured data to obtain labeled road test measured data; The laboratory standard energy consumption acquisition module 5 is used to predict the laboratory standard energy consumption of each typical road test area based on the marked road test data and the hub energy consumption prediction model. The energy consumption difference factor acquisition module 6 is used to calculate the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory. The energy consumption difference factor iterative optimization module 7 is used to perform correlation analysis between the energy consumption difference factor of the target new energy vehicle in each typical road test area and the corresponding regional characteristic parameters, and to iteratively optimize the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results. The regionalized coupled energy consumption prediction model construction module 8 is used to correct the hub energy consumption prediction model based on the iteratively optimized energy consumption difference factor and construct the regionalized coupled energy consumption prediction model. The coupled energy consumption prediction module is used to obtain the regional feature parameters of the area to be tested from the preset regional feature parameter library as the input of the regionalized coupled energy consumption prediction model, and to predict the coupled predicted energy consumption of the target new energy vehicle in the area to be tested.
[0053] The rest is the same as in Example 1.
[0054] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.
Claims
1. A method for predicting the regionalized energy consumption of new energy vehicles, characterized in that, Includes the following steps: In a standard laboratory environment, acquire basic energy consumption data for the target new energy vehicle; A hub energy consumption prediction model is trained based on the energy consumption data of the target new energy vehicle. Collect road test data of the target new energy vehicle in typical road test areas with significant differences; The road test data is discretized to obtain labeled road test data; Based on the road test data processed by the labeling, the laboratory standard energy consumption of each typical road test area is predicted by the hub energy consumption prediction model. The energy consumption difference factor of the target new energy vehicle in each typical road test area is calculated based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory. The energy consumption difference factors of the target new energy vehicle in each typical road test area are correlated with the corresponding regional characteristic parameters, and the energy consumption difference factors of the target new energy vehicle in each typical road test area are iteratively optimized based on the correlation analysis results. Based on the iteratively optimized energy consumption difference factor, the hub energy consumption prediction model is corrected to construct a regionalized coupled energy consumption prediction model. The regional feature parameters of the test area are obtained from the preset regional feature parameter library and used as the input of the regionalized coupled energy consumption prediction model to predict the coupled predicted energy consumption of the target new energy vehicle in the test area.
2. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The basic energy consumption data includes ambient set temperature, standard cycle operating condition energy consumption, and various vehicle selection modes.
3. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The road test data includes vehicle driving status, environmental data, and vehicle setting data. The vehicle driving status includes actual vehicle speed, energy consumption data, and vehicle CAN bus data. The environmental data includes temperature, humidity, slope, and altitude. The vehicle setting data includes vehicle mode, energy recovery mode, and air conditioning mode.
4. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The specific steps for discretizing the road test data to obtain labeled road test data include: The energy window method is adopted, and the continuous time series of road test measurement data is divided into a series of continuous energy windows with a preset energy value as the energy window unit; For each energy window, the corresponding environmental parameters, vehicle setting parameters, and driving status parameters are labeled to obtain labeled road test data.
5. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The hub energy consumption prediction model is trained using the RBF neural network method. The laboratory standard energy consumption for each road test area is predicted using this model. In the formula, For the target new energy vehicles in the first Laboratory standard energy consumption for each road test area To connect the first The weights of the radial basis functions to the output layer The total number of radial basis functions. For radial basis functions, In order to target the The labeled road test data vectors for each road test area. For the first The center vector of each radial basis function This is a bias term.
6. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The energy consumption difference factor is: In the formula, For the target new energy vehicles in the first Energy consumption difference factors in a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. For the target new energy vehicles in the first Laboratory standard energy consumption for a typical road test area.
7. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The regional characteristic parameters include climate and geographical parameters, traffic condition parameters, new energy vehicle user quantity parameters, and driving behavior parameters. The specific formula for iteratively optimizing the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results is as follows: In the formula, For the first The typical road test area in the first Optimize the energy consumption difference factor in round iteration. , and For dynamic weighting coefficients, For the first The typical road test area in the first The original energy consumption difference factor in the round of iteration, The moving average of the historical optimization energy consumption difference factor. For the first The typical road test area in the first The quantization correction term for the regional feature parameters in the round of iteration.
8. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The loss function of the regionalized coupled energy consumption prediction model is: In the formula, The loss function for the regionalized coupled energy consumption prediction model is... This represents the total number of typical road test areas. Weighting the mean square error between predicted and measured energy consumption. For the regionalized coupled energy consumption prediction model in the first Coupled predicted energy consumption for a typical road test area For the target new energy vehicles in the first Actual measured energy consumption during road tests in a typical road test area. The KL divergence weights represent the differences in the distribution of regional characteristics. Let KL divergence be the KL divergence. The empirical distribution of characteristic parameters of the region to be tested. For the first The baseline distribution of characteristic parameters of a typical test area. The final optimized Euclidean distance weight between the energy consumption difference factor and the baseline energy consumption difference factor is used. For the first The final optimized energy consumption difference factor for a typical road test area For the first Benchmark energy consumption difference factor for a typical road test area The regularization intensity coefficient is . For regularization terms of model parameters, The standard benchmark energy consumption for target new energy vehicles For regional feature coupling function, For the first A set of regional characteristic parameters for a typical road test area. These are the model parameters.
9. The regionalized energy consumption prediction method for new energy vehicles according to claim 1, characterized in that, The coupled predicted energy consumption is transformed into a regionalized energy consumption score through linear fitting, and the regionalized energy consumption score is as follows: In the formula, For the target new energy vehicle in the test area Regionalized energy consumption rating The coefficient representing the generally accepted increase in energy consumption during road tests compared to laboratory standard energy consumption. The benchmark score is 100 points. The coupled prediction of energy consumption of the target new energy vehicle in the test area.
10. A regionalized energy consumption prediction system for new energy vehicles, characterized in that, include: The energy consumption basic data acquisition module (1) is used to acquire the energy consumption basic data of the target new energy vehicle in a standard laboratory environment; The hub energy consumption prediction model training module (2) is used to train the hub energy consumption prediction model based on the energy consumption basic data of the target new energy vehicle. The road test data acquisition module (3) is used to collect road test data of the target new energy vehicle in a typical road test area with significant differences. The data labeling and processing module (4) is used to discretize the road test measured data to obtain the labeled road test measured data; The laboratory standard energy consumption acquisition module (5) is used to predict the laboratory standard energy consumption of each typical road test area based on the road test measured data processed by the marking and through the hub energy consumption prediction model. The energy consumption difference factor acquisition module (6) is used to calculate the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the measured energy consumption of the target new energy vehicle in each typical road test area and the standard energy consumption in the laboratory. The energy consumption difference factor iterative optimization module (7) is used to perform correlation analysis between the energy consumption difference factor of the target new energy vehicle in each typical road test area and the corresponding regional characteristic parameters, and to iteratively optimize the energy consumption difference factor of the target new energy vehicle in each typical road test area based on the correlation analysis results. The regionalized coupled energy consumption prediction model construction module (8) is used to correct the hub energy consumption prediction model based on the iteratively optimized energy consumption difference factor and construct the regionalized coupled energy consumption prediction model. The coupled energy consumption prediction module (9) is used to obtain the regional feature parameters of the area to be tested from the preset regional feature parameter library as the input of the regionalized coupled energy consumption prediction model, and predict the coupled predicted energy consumption of the target new energy vehicle in the area to be tested.