A method for constructing driving conditions of multi-feature pure electric vehicles
By optimizing the speed and battery signal data of pure electric vehicles and feature extraction, combined with the K-Means++ clustering method, a more accurate driving working condition was constructed, and the working condition caused by the failure to consider battery characteristics in the traditional method is accurately constructed. The problem of low degree improves the reliability of data quality and analysis results.
Patent Information
- Application Number
- CN202210611629.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-31
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-05-31
AI Technical Summary
The prior art does not fully consider the battery characteristics when building pure electric vehicles, resulting in poor accuracy in building operating conditions and rough data processing, which fails to reflect the characteristics of pure electric vehicles.
The driving working condition construction method of multi-feature pure electric vehicles is adopted. By acquiring and optimizing vehicle speed and battery signal data, data feature extraction and dimensionality reduction are performed, candidate working conditions are constructed using the K-Means++ clustering method, and feature parameters are calculated to select the minimum error. Representative working condition.
It improves the accuracy and data quality of the driving conditions analysis results of pure electric vehicles, enriches the working conditions construction theory, and significantly reduces the relative error between the working conditions and the overall data.
Smart Images

Figure CN114970730B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of automobile driving performance testing, and in particular to a method for constructing a multi-feature pure electric vehicle driving condition. Background Art
[0002] Automobile operating conditions refer to the working conditions of automobiles during transportation and driving. According to the movement form of the automobile, there are mainly driving conditions such as starting, accelerating, constant speed, decelerating, turning, uphill and downhill, and parking. Automobile driving conditions are also called vehicle test cycles, which are speed-time curves that describe automobile driving. On the basis of collecting a large amount of measured data, constructing actual road cycle conditions is the data support for vehicle optimization testing and simulation. Automobile driving conditions are an important basic standard in the automotive industry, an important design input in the process of automobile product development, the main benchmark for calibration and optimization of various performance indicators of automobiles, and an important basis for evaluating automobile emissions, fuel consumption, control strategies and parameter matching. They are common core technologies in automobile research.
[0003] In recent years, with the depletion of global oil resources, new energy vehicle technology has developed rapidly, especially pure electric vehicle technology has made major breakthroughs, both the battery life and the safety performance of the whole vehicle have been effectively improved. With the development of pure electric vehicle technology, the research on pure electric vehicle driving conditions has gradually been paid attention to. Now the short-stroke method is generally used to decompose the motion fragments to study the driving conditions. Although this method has also achieved certain research results in driving conditions, there are still many problems. For example, the current construction of pure electric vehicle driving conditions still uses the characteristics of traditional fuel vehicles to construct conditions, and no further mining of the characteristics is carried out. The power sources of pure electric vehicles and traditional fuel vehicles are different, and the characteristics of pure electric vehicle batteries are not considered in the characteristic parameters, which will affect the accuracy of the construction of pure electric vehicle conditions. Only some outliers are deleted from the original data, and the data details are not deeply investigated, which leads to insufficient feature parameters of the synthetic candidate conditions, which do not reflect the characteristics of pure electric vehicles, and the processing of the original data is relatively rough, which ultimately leads to poor accuracy in the construction of pure electric vehicle conditions. Summary of the invention
[0004] The present invention aims to provide a method for constructing a multi-feature pure electric vehicle driving condition to improve the accuracy of the pure electric vehicle driving condition analysis results.
[0005] To achieve the above object, the present invention adopts the following technical solution: a method for constructing a multi-feature pure electric vehicle driving condition, comprising the following steps:
[0006] Step S1, obtaining the operation data of the pure electric vehicle and optimizing the operation data to obtain a first data set, and dividing the first data set into a plurality of movement segments;
[0007] Step S2, optimizing data for each motion segment, and then extracting data features from the optimized motion segments from multiple dimensions;
[0008] Step S3, using principal component analysis to reduce the dimension of the data features of the motion segments, and using K-Means++ clustering method to perform unsupervised cluster analysis on the reduced-dimensional motion segments, defining the speed range according to the speed features of each category in the clustering results, and after setting the total duration of the working condition, selecting multiple motion segments near the cluster center to construct multiple candidate working conditions;
[0009] Step S4, calculate the characteristic parameters of the candidate working conditions that do not contain extreme values, calculate the relative error value with the overall data, and select the working condition with the smallest average error as the representative working condition output.
[0010] The principles and advantages of this solution are as follows: in actual application, the original message data of the target pure electric vehicle is first parsed to obtain the corresponding vehicle speed signal data and battery signal data, and then the vehicle speed signal data and battery signal data are cleaned and optimized to improve the reliability of the data. A mean filter is introduced into the battery signal in each motion segment for smoothing to eliminate the instability of the battery signal and undiscovered sampling anomalies. At the same time, the principal component analysis method is used to reduce the dimension of the motion segment features, and the K-Means++ clustering method is used to perform unsupervised clustering analysis on the reduced motion segments. The speed range is defined according to the speed characteristics of each category in the clustering results, and the total working condition duration is set. The motion segments that are close to the centroid and meet the duration requirements are selected for combination and splicing to construct multiple candidate working conditions. Finally, the characteristic parameters that do not contain extreme values in the candidate working conditions are calculated, and the relative error value is calculated with the overall data. The working condition with the smallest average error is selected as the representative working condition output.
[0011] Compared with the prior art, the advantages of the present invention lie in that it extracts the battery characteristics of pure electric vehicles, enriches the feature library, and takes into account the instability of battery signals to perform secondary processing on battery current data, further improving data quality, effectively improving the data quality and representativeness of data fragments of pure electric vehicle driving conditions, significantly reducing the relative error between the operating conditions and the overall data, greatly improving the accuracy of driving condition analysis results, and enriching the theory of pure electric vehicle operating condition construction.
[0012] Preferably, as an improvement, the operating data includes vehicle speed signal data and battery signal data, and the operating data is optimized by cleaning the vehicle speed and battery signal data, deleting obviously abnormal data, and using linear interpolation to fill in the missing one second of data.
[0013] Beneficial effects: By collecting vehicle speed data and battery data, the diversity of working condition construction features is increased, the reliability of the data is improved, and at the same time, the data is optimized and processed, invalid data is eliminated, and the validity of the data is improved, which facilitates the subsequent construction of driving conditions and improves the accuracy of the working condition output results.
[0014] Preferably, as an improvement, the obviously abnormal data includes data of time repetition, abnormal vehicle speed, abnormal longitude and latitude, and abnormal acceleration.
[0015] Beneficial effects: Considering the complexity of driving conditions, there will be a lot of abnormal data in the collected data. Therefore, the abnormal data should be cleaned up before data processing and calculation to reduce the amount of calculation of subsequent data, reduce the difficulty of data refinement processing, improve data processing efficiency, and effectively improve the accuracy of calculation results.
[0016] Preferably, as an improvement, the first data set is divided into multiple motion segments, the rate of change of the battery current is calculated, and the cleaned data is divided into multiple motion segments according to the battery current value characteristics and the rate of change of the battery current; the motion segments include idle motion segments, acceleration motion segments, uniform motion segments and deceleration motion segments.
[0017] Beneficial effects: According to the characteristics of battery signal changes during actual driving of pure electric vehicles, the motion segments are divided into the above four types, which facilitates the unified division of vehicle speed data and improves the accuracy of data analysis in the analysis of subsequent driving conditions, thereby ensuring the accuracy of the output conditions.
[0018] Preferably, as an improvement, data optimization is performed on each motion segment by introducing a mean filter into the battery signal in each motion segment for smoothing to eliminate the instability of the battery signal and undetected sampling anomalies. The motion segment is judged based on the characteristics of the smoothed current value to determine whether there is a sampling anomaly. If the current change is too large, the motion segment is eliminated.
[0019] Beneficial effect: Optimizing the battery signal data in the motion segment can effectively improve the quality of the battery signal data, thereby improving the reliability and accuracy of the subsequent operating condition analysis results, and ensuring the most accurate operating condition information of the pure electric vehicle.
[0020] Preferably, as an improvement, the data characteristics include time characteristics, motion characteristics and battery characteristics; the time characteristics include the duration of the motion segment, the idle time ratio, the acceleration time ratio, the uniform speed time ratio, the deceleration time ratio, the battery discharge time ratio and the battery energy recovery time ratio; the motion characteristics include the maximum speed, the average speed, the average driving speed, the speed standard deviation, the maximum acceleration, the average acceleration, the acceleration standard deviation, the maximum deceleration, the average deceleration and the deceleration standard deviation; the battery characteristics include the maximum discharge current, the average discharge current, the discharge current standard deviation, the maximum energy recovery current, the energy recovery average current, the energy recovery current standard deviation, the discharge current maximum acceleration, the discharge current average acceleration, the discharge current acceleration standard deviation, the energy recovery current maximum deceleration, the energy recovery current average deceleration and the energy recovery current deceleration standard deviation.
[0021] Beneficial effects: Selecting the above characteristics to calculate and analyze the driving conditions of pure electric vehicles not only fully considers the characteristics of pure electric vehicle batteries, but also enriches the database, improves the diversity of data and the reliability of analysis results, thereby ensuring the accuracy of the working condition output results.
[0022] Preferably, as an improvement, when performing feature dimensionality reduction on the data features of the motion segment, a plurality of principal components whose contribution rates are greater than a first threshold are selected for subsequent analysis and calculation.
[0023] Beneficial effects: Feature dimensionality reduction of motion segment features can effectively reduce the amount of data calculation, eliminate data correlation, and improve the accuracy of driving condition analysis results. At the same time, by limiting the contribution rate threshold and the number of principal components, the amount of data can be effectively eliminated and the analysis and calculation speed can be improved.
[0024] Preferably, as an improvement, the first threshold is 85%; and the plurality of principal components are the top five principal components.
[0025] Beneficial effect: By calculating this threshold based on the variance, the type of principal component for subsequent analysis and calculation can be accurately limited. While ensuring data reliability, the amount of data can be streamlined, the analysis speed of driving conditions can be improved, and the first five principal components that meet the requirements can be selected at the same time. While ensuring the accuracy of the calculation results, the amount of calculation can be effectively reduced.
[0026] Preferably, as an improvement, the speed interval includes a low speed interval, a medium speed interval and a high speed interval.
[0027] Beneficial effect: Dividing the speed intervals according to speed characteristics can not only ensure the integrity of the motion segments, but also make the speed interval division more in line with the actual working conditions, thereby ensuring the accuracy of the subsequent output working condition analysis results.
[0028] Preferably, as an improvement, when multiple motion clips are screened from the vicinity of the cluster center to construct multiple candidate operating conditions, motion clips that are closer to the centroid and meet the duration requirement are selected for combination and splicing, and finally multiple candidate driving conditions are constructed.
[0029] Beneficial effect: By selecting motion segments near the cluster center for combination and splicing, the accuracy of the constructed driving condition data can be ensured, and when the error calculation of the condition is performed subsequently, the calculation speed and the accuracy of the calculation results can be effectively guaranteed, thereby ensuring the accuracy of the obtained driving condition. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a flow chart of Embodiment 1 of a method for constructing driving conditions of a multi-feature pure electric vehicle according to the present invention.
[0031] Figure 2 A schematic diagram of a driving condition of a first embodiment of a method for constructing a driving condition of a multi-feature pure electric vehicle according to the present invention. DETAILED DESCRIPTION
[0032] The following is further described in detail through specific implementation methods:
[0033] Embodiment 1:
[0034] This embodiment is basically as shown in the attached Figure 1 As shown: A method for constructing a multi-feature pure electric vehicle driving condition includes the following steps:
[0035] Step S1, obtaining the operation data of the pure electric vehicle and optimizing the operation data to obtain a first data set, and dividing the first data set into a plurality of movement segments;
[0036] Step S2, optimizing data for each motion segment, and then extracting data features from the optimized motion segments from multiple dimensions;
[0037] Step S3, using principal component analysis to reduce the dimension of the data features of the motion segments, and using K-Means++ clustering method to perform unsupervised cluster analysis on the reduced-dimensional motion segments, defining the speed range according to the speed features of each category in the clustering results, and after setting the total duration of the working condition, selecting multiple motion segments near the cluster center to construct multiple candidate working conditions;
[0038] Step S4, calculate the characteristic parameters of the candidate working conditions that do not contain extreme values, calculate the relative error value with the overall data, and select the working condition with the smallest average error as the representative working condition output.
[0039] Specifically, in step S1, the operating data includes vehicle speed signal data and battery signal data. The operating data is optimized by cleaning the vehicle speed and battery signal data, deleting obviously abnormal data, including time duplication, abnormal vehicle speed, abnormal latitude and longitude, and abnormal acceleration data, and using linear interpolation to fill in the missing one second of data. Then, the short-stroke method is used to record each idle starting point with a speed of 0 as a segmentation point, and the rate of change of the battery current is calculated. According to the characteristics of the battery current value and the rate of change of the battery current, the cleaned data is divided into multiple motion segments, including idle motion segments, acceleration motion segments, uniform speed motion segments, and deceleration motion segments.
[0040] Specifically, in step S2, when optimizing the data for each motion segment, a mean filter is introduced into the battery signal in each motion segment for smoothing to eliminate the instability of the battery signal and undiscovered sampling anomalies. The sliding window size of the mean filter is 5, and it slides 1 step each time to calculate the current mean of the battery in the sliding window. After the sliding window scans a motion segment, the current change is checked, and it is judged whether there is a sampling anomaly in this motion segment based on the smoothed current value characteristics. If there is an abnormal mutation or a steep current curve in the smoothed current curve, it is considered that there is a sampling anomaly in the motion segment, so this motion segment is eliminated.
[0041] Then, the optimized motion segments are feature extracted from three dimensions: time feature, motion feature, and battery feature. The results are shown in Table 1. The time features include motion segment duration, idle time ratio, acceleration time ratio, uniform speed time ratio, deceleration time ratio, battery discharge time ratio, and battery energy recovery time ratio.
[0042] The motion characteristics include maximum speed, average speed, average driving speed, speed standard deviation, maximum acceleration, average acceleration, acceleration standard deviation, maximum deceleration, average deceleration, and deceleration standard deviation;
[0043] The battery characteristics include maximum discharge current, average discharge current, standard deviation of discharge current, maximum energy recovery current, average energy recovery current, standard deviation of energy recovery current, maximum acceleration of discharge current, average acceleration of discharge current, standard deviation of discharge current acceleration, maximum deceleration of energy recovery current, average deceleration of energy recovery current and standard deviation of energy recovery current deceleration.
[0044] Table 1: Motion clip data feature table
[0045]
[0046]
[0047] Specifically, in step S3, after the principal component analysis method is used to reduce the dimension of the data features of the motion clip, the first five principal components with a cumulative contribution rate greater than 85% are selected for subsequent analysis and calculation. The principal component analysis results are shown in Table 2. By limiting the contribution rate and the number of principal components, the data volume can be streamlined while ensuring the validity of the data to the greatest extent, thereby reducing the amount of other invalid and insignificant data, reducing the difficulty of data analysis, and improving the reliability of data analysis results.
[0048] Table 2: Variance contribution of each principal component
[0049] principal component P1 P2 P3 P4 P5 Variance contribution rate (%) 35.25 25.25 12.05 8.95 6.66 Cumulative contribution rate (%) 35.26 60.51 72.56 81.51 88.17
[0050] Then, the K-Means++ clustering method is used to perform unsupervised cluster analysis on the motion segments after dimensionality reduction. Then, three speed intervals of low speed, medium speed and high speed are defined according to the speed characteristics of each category in the clustering results. Considering the integrity of the kinematic segments and referring to the NEDC, WLTC and China CATC conditions, the driving condition duration of pure electric vehicles is limited to the duration range of various cluster segments of 1800 seconds. The calculation formula is as follows:
[0051]
[0052] Where, T k is the duration of the k-th motion segment in the candidate working condition, T total is the total duration of the final working condition, t i,k is the duration of the i-th segment of the k-th motion segment.
[0053] The time length range corresponding to each interval is calculated, low-speed interval: 250s, medium-speed interval: 644s, high-speed interval: 906s. At the same time, in order to avoid selecting motion clips with ambiguous categories, when splicing motion clips of each category, the clips that are closer to the center of mass and meet the duration requirements are selected for combined splicing, and a total of 4 candidate working conditions are constructed.
[0054] Finally, the characteristic parameters that do not contain extreme values in the four candidate working conditions are calculated. The results are shown in Table 3. The relative error value is calculated with the total data of the four optimized intervals. The working condition with the smallest average error is selected as the representative working condition output. The output representative working condition is shown in the attached figure. Figure 2 shown.
[0055] Table 3: Candidate operating conditions results table
[0056]
[0057]
[0058] The specific implementation process of this embodiment is as follows:
[0059] In the first step, the GB32960 national standard original message data of the pure electric vehicle is parsed to obtain the corresponding vehicle speed signal data and battery signal data, and the acquired vehicle speed signal data and battery signal data are cleaned to delete the obviously abnormal data, including time duplication, abnormal vehicle speed, abnormal latitude and longitude, and abnormal acceleration data, and linear interpolation is used to fill the missing one second of data. Then, the short-stroke method is used to record each idle starting point with a speed of 0 as a segmentation point, and the rate of change of the battery current is calculated. According to the characteristics of the battery current value and the rate of change of the battery current, the processed vehicle speed signal data is divided into idle motion segments, acceleration motion segments, uniform motion segments, and deceleration motion segments.
[0060] In the second step, a mean filter is introduced into each battery signal in each motion segment for smoothing to eliminate the instability of the battery signal and undiscovered sampling anomalies. The sliding window size of the mean filter is 5, and it slides 1 step each time to calculate the current mean of the battery in the sliding window. After the sliding window scans a motion segment, the current change is checked, and the sampling anomaly of this motion segment is judged based on the characteristics of the smoothed current value. If the current curve has an abnormal mutation or a steep current curve after smoothing, it is considered that there is a sampling anomaly in the motion segment, so this motion segment is eliminated.
[0061] The third step is to extract features of the optimized motion segments from three dimensions: time features, motion features and battery features. The time features include the duration of the motion segments, the proportion of idling time, the proportion of acceleration time, the proportion of uniform speed time, the proportion of deceleration time, the proportion of battery discharge time and the proportion of battery energy recovery time; the motion features include the maximum speed, average speed, average driving speed, speed standard deviation, maximum acceleration, average acceleration, acceleration standard deviation, maximum deceleration, average deceleration and deceleration standard deviation; the battery features include the maximum discharge current, average discharge current, discharge current standard deviation, maximum energy recovery current, average energy recovery current, energy recovery current standard deviation, maximum acceleration of discharge current, average acceleration of discharge current, standard deviation of discharge current acceleration, maximum deceleration of energy recovery current, average deceleration of energy recovery current and standard deviation of deceleration of energy recovery current.
[0062] In the fourth step, the principal component analysis method is used to reduce the dimension of the motion segment features, and the top five principal components with a cumulative contribution rate greater than 85% are selected for subsequent calculations. Then, the K-Means++ clustering method is used to perform unsupervised clustering analysis on the motion segments after dimensionality reduction. Then, according to the speed characteristics of each category in the clustering results, three speed intervals of low speed, medium speed, and high speed are defined. Considering the integrity of the kinematic segments and referring to the NEDC, WLTC and China CATC conditions, the driving condition duration of pure electric vehicles is limited to 1800 seconds for each type of cluster segment duration range, and the corresponding time length range for each interval is calculated according to the corresponding formula, among which, the low-speed interval: 250s, the medium-speed interval: 644s, and the high-speed interval: 906s. At the same time, in order to avoid selecting motion segments with ambiguous categories, when splicing each type of motion segments, the segments that meet the duration requirements and are closer to the center of mass are selected for combined splicing, and a total of 4 candidate conditions are constructed.
[0063] The fifth step is to calculate the characteristic parameters that do not contain extreme values in the four candidate working conditions, and calculate the relative error value with the total data of the four optimized intervals, and select the working condition with the smallest average error as the representative working condition output.
[0064] The automobile driving condition is an important basic standard in the automobile industry, an important design input in the process of automobile product development, the main benchmark for calibration and optimization of various automobile performance indicators, and an important basis for evaluating automobile emissions, fuel consumption, control strategies and parameter matching. It is a common core technology in automobile research. Therefore, with the development of pure electric vehicle technology and the frequent accidents of pure electric vehicles in recent years, the research on pure electric vehicle driving conditions has gradually been paid attention to. At present, the construction of pure electric vehicle driving conditions still uses the characteristics of traditional fuel vehicles to construct conditions, and has not been deeply mined. In fact, the power sources of pure electric vehicles and traditional fuel vehicles are different. If the characteristics of pure electric vehicle batteries are not considered in the characteristic parameters, it will affect the accuracy of the construction of pure electric vehicle conditions. Only some outliers are deleted from the original data without going into the data details, which leads to insufficient feature parameters of the synthetic candidate conditions, which do not reflect the characteristics of pure electric vehicles, and the processing of the original data is relatively rough, which ultimately leads to poor accuracy in the construction of pure electric vehicle conditions and cannot reflect the real driving conditions of pure electric vehicles.
[0065] This solution focuses on the research and analysis of the driving conditions of pure electric vehicles, and focuses on overcoming the technical bias of the main research characteristics of driving conditions in the conventional concept, and focuses on extracting the battery characteristics of pure electric vehicles. The starting point is unexpected. Taking into account the instability of battery signals, the battery current data is processed twice, which further improves the data quality, effectively improves the data quality and data segment representativeness of the driving conditions of pure electric vehicles, and the entire analysis idea is non-obvious. Specifically, the driving conditions of pure electric vehicles are constructed based on the improved short-stroke method. For the vehicle speed signal data, the division of motion segments is mainly improved. Each idle starting point with a speed of 0 is recorded as a segmentation point, and the processed data is divided into multiple motion segments, including idle motion segments, acceleration motion segments, uniform speed motion segments and deceleration motion segments, thereby increasing the number of optional motion segments. Combined with the extracted battery characteristics, it not only enriches the feature database, but also further improves the data quality and the representativeness of data segments due to the optimization of the battery signal data.
[0066] Finally, the motion segments after data optimization are subjected to feature extraction and dimensionality reduction analysis from three dimensions: time features, motion features, and battery features. The segments near the center of mass that meet the duration requirements are selected for combination and splicing to construct multiple candidate operating conditions. Finally, the error of each candidate operating condition is calculated based on the total data of the four optimized intervals, and the candidate operating condition with the smallest error rate is selected as the most representative operating condition output, which significantly reduces the relative error between the synthetic operating condition and the overall data, ensures the accuracy of the pure electric vehicle driving condition analysis results, and provides real and reliable reference data for the study of vehicle performance, and enriches the theory of pure electric vehicle operating condition construction.
[0067] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme is not described in detail here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The protection scope required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.
Claims
1. A method for constructing a multi-feature pure electric vehicle driving condition, characterized in that: The following steps are involved: Step S1, obtaining the operation data of the pure electric vehicle and optimizing the operation data to obtain a first data set, and dividing the first data set into a plurality of movement segments; Step S2, optimizing data for each motion segment, and then extracting data features from the optimized motion segments from multiple dimensions; Step S3, using principal component analysis to reduce the dimension of the data features of the motion segments, and using K-Means++ clustering method to perform unsupervised cluster analysis on the reduced-dimensional motion segments, defining the speed range according to the speed features of each category in the clustering results, and after setting the total duration of the working condition, selecting multiple motion segments near the cluster center to construct multiple candidate working conditions; Step S4, calculating the characteristic parameters of the candidate working conditions that do not contain extreme values, and calculating the relative error value with the overall data, and selecting the working condition with the smallest average error as the representative working condition output; The operation data includes vehicle speed signal data and battery signal data; the optimization processing of the operation data includes cleaning the vehicle speed and battery signal data, deleting obviously abnormal data, using linear interpolation to fill in the missing one second of data, and using the short stroke method to record each idle start point with a speed of 0 as a segmentation point, and calculating the change rate of the battery current, and dividing the cleaned data into multiple motion segments according to the battery current value characteristics and the battery current change rate; Parsing the corresponding vehicle speed signal data and battery signal data from the original message data of the target pure electric vehicle; The data optimization for each motion segment is to introduce a mean filter into the battery signal in each motion segment to perform smoothing processing to eliminate the instability of the battery signal and undiscovered sampling anomalies. The sliding window size of the mean filter is 5, and each sliding step is 1 step. The current mean of the battery in the sliding window is calculated. After the sliding window scans a motion segment, the current change is checked. According to the smoothed current value characteristics, it is determined whether there is a sampling anomaly in this motion segment. If the current change is too large, this motion segment is eliminated. The data features include time features, motion features and battery features; the time features include the duration of motion segments, idle time ratio, acceleration time ratio, uniform speed time ratio, deceleration time ratio, battery discharge time ratio and battery energy recovery time ratio; the motion features include maximum speed, average speed, average driving speed, speed standard deviation, maximum acceleration, average acceleration, acceleration standard deviation, maximum deceleration, average deceleration and deceleration standard deviation; the battery features include maximum discharge current, average discharge current, discharge current standard deviation, maximum energy recovery current, average energy recovery current, energy recovery current standard deviation, maximum discharge current acceleration, average discharge current acceleration, discharge current acceleration standard deviation, maximum energy recovery current deceleration, average energy recovery current deceleration and energy recovery current deceleration standard deviation.
2. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 1, characterized in that: The obviously abnormal data include data of time duplication, abnormal vehicle speed, abnormal longitude and latitude, and abnormal acceleration.
3. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 1, characterized in that: The first data set is divided into multiple motion segments by calculating the rate of change of the battery current, and dividing the cleaned data into multiple motion segments according to the battery current value characteristics and the rate of change of the battery current; the motion segments include idle motion segments, acceleration motion segments, uniform motion segments and deceleration motion segments.
4. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 1, characterized in that: When performing feature dimensionality reduction on the data features of the motion segment, a plurality of principal components whose contribution rates are greater than a first threshold are selected for subsequent analysis and calculation.
5. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 4, characterized in that: The first threshold is 85%; the plurality of principal components are the top five principal components.
6. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 1, characterized in that: The speed intervals include a low speed interval, a medium speed interval and a high speed interval.
7. A method for constructing a multi-feature pure electric vehicle driving condition according to claim 1, characterized in that: When selecting multiple motion segments near the cluster center to construct multiple candidate operating conditions, motion segments that are closer to the centroid and meet the duration requirement are selected for combination and splicing, and finally multiple candidate driving operating conditions are constructed.
Citation Information
Patent Citations
Pure electric vehicle drive working condition recognizing method based on multi-information fusion
CN104176058A
Novel electric vehicle driving state construction method
CN110866997A
Pure electric vehicle driving condition construction method
CN113297795A
Vehicle driving condition construction method based on improved short-stroke method
CN114943296A
Detection method and apparatus, storage medium
WO2021164463A1
