Search

CN-121997204-A - Intelligent diagnosis analysis method suitable for energy consumption abnormality of new energy vehicle

CN121997204ACN 121997204 ACN121997204 ACN 121997204ACN-121997204-A

Abstract

The invention relates to the technical field of energy consumption analysis of new energy vehicles, in particular to an intelligent diagnosis analysis method suitable for energy consumption abnormality of a new energy vehicle, which comprises the following steps of collecting cloud big data historical data of a vehicle network, wherein the data comprise CAN bus data, GPS (global positioning system) position information, vehicle type and customer information, electric accessory operation parameters and air conditioner operation parameters of the vehicle; the method comprises the steps of preprocessing collected data, extracting multidimensional high-reliability characteristic factors affecting energy consumption by a design algorithm, carrying out anomaly detection on monthly hundred kilometers of energy consumption values of the same vehicle type or the same order vehicle based on a box diagram method of a statistical principle, identifying abnormal points and generating a data label. The system and the method have comprehensive data acquisition, cover various data such as vehicle operation, charging and the like, ensure the comprehensiveness of analysis, provide rich input for SHAP value calculation, extract characteristics in multiple dimensions, cover a plurality of influencing factors such as vehicle performance, operation conditions, driving behaviors and the like, improve the accuracy of abnormality diagnosis and ensure the reliability of SHAP value calculation.

Inventors

  • Shao Kongmu
  • REN YONGHUAN
  • ZHENG BINBIN
  • LIN JIAXIANG
  • SU LIANG

Assignees

  • 厦门金龙联合汽车工业有限公司

Dates

Publication Date
20260508
Application Date
20260128

Claims (10)

  1. 1. The intelligent diagnosis and analysis method for the abnormal energy consumption of the new energy vehicle is characterized by comprising the following steps of: s1, acquiring cloud big data historical data of a vehicle networking, wherein the data comprise CAN bus data, GPS position information, vehicle type information, customer information, electric accessory operation parameters and air conditioner operation parameters of a vehicle; S2, preprocessing the data acquired in the step S1, and extracting multidimensional high-reliability characteristic factors affecting energy consumption by a design algorithm; S3, a box diagram method based on a statistical principle is used for carrying out anomaly detection on the monthly hundred kilometer energy consumption value of the same vehicle type or the same order (the same batch as a customer) of the vehicle, identifying an anomaly point and generating a data tag; S4, constructing a data set, performing training fitting on the data set by adopting a machine learning method based on a tree model, and outputting an abnormal probability value of the energy consumption of the same vehicle type or the same order (same batch as the client); s5, initializing TreeExplainer an interpreter, calculating SHAP values of all features of each abnormal vehicle based on a Tree SHAP algorithm, and determining the feature with the largest influence; s6, converting and calculating SHAP values to obtain influence weight indexes of the features; S7, outputting an abnormal vehicle number, abnormal probability, influence maximum characteristics and influence weights.
  2. 2. The intelligent diagnosis and analysis method for abnormal energy consumption of new energy vehicles according to claim 1, wherein in step S1, the CAN bus data includes mileage data, vehicle speed data, battery voltage data, battery current data, motor operation parameters, and transmission system operation parameters.
  3. 3. The intelligent diagnosis and analysis method for abnormal energy consumption of a new energy vehicle according to claim 1, wherein in step S2, the multidimensional high reliability characteristic factors include motor efficiency, transmission efficiency, road gradient, air conditioner energy consumption ratio, electric accessory energy consumption ratio, charging current, recovery efficiency, charging process water cooling unit energy consumption ratio, acceleration characteristic, high speed ratio, idle speed ratio and parking air conditioner on recognition ratio.
  4. 4. The intelligent diagnosis analysis method for the abnormal energy consumption of the new energy vehicle is characterized in that the calculation mode of the high-speed duty ratio is that the numerical number N1 of the vehicle speed greater than 50km/h is extracted and divided by the numerical number N2 of all the vehicle speeds other than 0 to obtain the high-speed duty ratio=N1/N2, and the calculation mode of the idle speed duty ratio is that the idle speed total time of the vehicle is counted and divided by the vehicle power-on total time to obtain the idle speed duty ratio=idle speed total time/power-on total time.
  5. 5. The intelligent diagnosis analysis method for the abnormal energy consumption of the new energy vehicle is characterized in that the recovery efficiency is calculated by identifying a non-charging section of the vehicle, counting the power value product of total voltage and current, calculating the absolute value E1 of negative power summation and the positive power summation E2, and the recovery efficiency=E1/E2, wherein the charging current is determined by identifying a charging time starting point and a charging time ending point, collecting current values in the charging process, and calculating the charging current through probability density distribution.
  6. 6. The intelligent diagnosis analysis method for abnormal energy consumption of a new energy vehicle according to claim 3 is characterized in that the energy consumption in the step S2 comprises battery thermal management energy consumption, the extraction mode of relevant characteristics of battery thermal management energy consumption is that when the vehicle adopts an independent water cooling mode, a TMS required power value is counted and integral operation is carried out, the calculation mode of a parking air conditioner opening recognition ratio is that the time period occupation ratio of the air conditioner which is not closed when the vehicle is parked is counted, and the parking air conditioner opening recognition ratio is determined by combining the use habit of the vehicle.
  7. 7. The intelligent diagnosis and analysis method for abnormal energy consumption of a new energy vehicle according to claim 1, wherein the abnormality detection process of the box-line diagram method in step S3 comprises: s31, calculating a first quartile Q1 and a third quartile Q3 of the same batch of vehicle energy consumption data; s32, calculating a quartile range IQR=Q3-Q1; S33, determining a critical value, wherein the lower limit critical value=Q1-1.5×IQR, and the upper limit critical value=Q3+1.5×IQR; S34, judging whether the energy consumption value of hundred kilometers per month is higher than an upper limit critical value, if so, marking the energy consumption value as the upper limit abnormal value to generate an abnormal label, and if not, marking the energy consumption value as a normal label.
  8. 8. The intelligent diagnosis and analysis method for abnormal energy consumption of new energy vehicles according to claim 1, wherein in step S4, the machine learning method based on the tree model is a random forest classifier, and the training process of the random forest classifier comprises data set division, super-parameter tuning and model verification.
  9. 9. The intelligent diagnosis and analysis method for abnormal energy consumption of new energy vehicles according to claim 1, wherein in step S5, the SHAP value is calculated by the following formula: Determining the feature with the greatest influence, wherein phi is the SHAP value of the feature i, F is the set of all the features, S is the feature subset, F (S) is the marginal contribution of the subset S, and TreeExplainer the interpreter directly calculates the feature contribution through the tree structure feature, wherein the tree structure feature comprises the node splitting condition and the leaf node weight, and the accurate Shapley value can be obtained without sampling.
  10. 10. The method for intelligently diagnosing and analyzing the abnormal energy consumption of the new energy vehicle according to claim 1, wherein in the step S6, the influence weight index is calculated by dividing a single characteristic SHAP value of an abnormal sample by the sum of all characteristic SHAP values of the abnormal sample to obtain the influence weight index corresponding to the characteristic.

Description

Intelligent diagnosis analysis method suitable for energy consumption abnormality of new energy vehicle Technical Field The invention relates to the technical field of energy consumption analysis of new energy vehicles, in particular to an intelligent diagnosis analysis method suitable for energy consumption abnormality of a new energy vehicle. Background Under the background of green transformation and 'double carbon' strategic deepening promotion in the traffic field, new energy buses become core carriers of urban and inter-city green public transportation, and the energy consumption efficiency is directly related to cost control, energy conservation and emission reduction achievements of operators. With the rapid expansion of new energy automobile markets and the continuous increase of vehicle conservation, the monitoring and analysis requirements for abnormal vehicle energy consumption in the industry are increasingly urgent. At present, monitoring and analysis of energy consumption abnormality of new energy vehicles in industry mostly depends on experience judgment and post manual investigation of operation and maintenance personnel based on limited data. This conventional mode has a number of drawbacks: 1. The analysis granularity is coarse, and abnormal energy consumption caused by multiple factors cannot be accurately captured; 2. The timeliness is poor, and the requirement of real-time energy consumption management is difficult to meet in post-investigation; 3. The system is highly dependent on manpower, and cannot cope with real-time and high-frequency data generated by mass vehicles, so that large-scale, universalization and fine energy consumption management are difficult to realize. In the prior related patent technologies, partial schemes aim at abnormal analysis of energy consumption of the fuel oil vehicles and cannot adapt to the energy consumption characteristics of new energy vehicles, the partial schemes are only limited to analysis of specific fault reasons or single influencing factors, the application range is narrow, and the partial schemes adopt a simple statistical method for diagnosis, so that the accuracy and the efficiency are low. Therefore, an intelligent analysis method capable of accurately and rapidly identifying the abnormal energy consumption of the new energy vehicle and automatically attributing to the abnormal energy consumption is urgently needed to solve the defects in the prior art. Disclosure of Invention Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and the appended drawings. The invention aims to overcome the defects and provide the intelligent diagnosis and analysis method for the abnormal energy consumption of the new energy vehicle. In order to achieve the purpose, the technical scheme of the invention is that the intelligent diagnosis and analysis method for the energy consumption abnormality of the new energy vehicle comprises the following steps: s1, acquiring cloud big data historical data of a vehicle networking, wherein the data comprise CAN bus data, GPS position information, vehicle type information, customer information, electric accessory operation parameters and air conditioner operation parameters of a vehicle; S2, preprocessing the data acquired in the step S1, and extracting multidimensional high-reliability characteristic factors affecting energy consumption by a design algorithm; S3, a box diagram method based on a statistical principle is used for carrying out anomaly detection on the monthly hundred kilometer energy consumption value of the same vehicle type or the same order (the same batch as a customer) of the vehicle, identifying an anomaly point and generating a data tag; S4, constructing a data set, performing training fitting on the data set by adopting a machine learning method based on a tree model, and outputting an abnormal probability value of the energy consumption of the same vehicle type or the same order (same batch as the client); s5, initializing TreeExplainer an interpreter, calculating SHAP values of all features of each abnormal vehicle based on a Tree SHAP algorithm, and determining the feature with the largest influence; s6, converting and calculating SHAP values to obtain influence weight indexes of the features; S7, outputting an abnormal vehicle number, abnormal probability, influence maximum characteristics and influence weights. In some embodiments, in step S1, the CAN bus data includes mileage data, vehicle speed data, battery voltage data, battery current data, motor operating parameters, driveline operating parameters. In some embodiments, in step S2, the multi-dimensional high reliability feature factors inclu