CN114118217A - A battery insulation failure prediction method for electric vehicles - Google Patents

A battery insulation failure prediction method for electric vehicles Download PDF

Info

Publication number
CN114118217A
CN114118217A CN202111263384.2A CN202111263384A CN114118217A CN 114118217 A CN114118217 A CN 114118217A CN 202111263384 A CN202111263384 A CN 202111263384A CN 114118217 A CN114118217 A CN 114118217A
Authority
CN
China
Prior art keywords
data
insulation failure
battery insulation
battery
machine learning
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202111263384.2A
Other languages
Chinese (zh)
Inventor
丁磊
韩大鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Human Horizons Shanghai Internet Technology Co Ltd
Original Assignee
Human Horizons Shanghai Internet Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Human Horizons Shanghai Internet Technology Co Ltd filed Critical Human Horizons Shanghai Internet Technology Co Ltd
Priority to CN202111263384.2A priority Critical patent/CN114118217A/en
Publication of CN114118217A publication Critical patent/CN114118217A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21—Design, administration or maintenance of databases
    • G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/25—Integrating or interfacing systems involving database management systems
    • G06F16/254—Extract, transform and load [ETL] procedures, e.g. ETL data flows in data warehouses
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/28—Databases characterised by their database models, e.g. relational or object models
    • G06F16/284—Relational databases
    • G06F16/285—Clustering or classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/245—Classification techniques relating to the decision surface
    • G06F18/2451—Classification techniques relating to the decision surface linear, e.g. hyperplane
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Business, Economics & Management (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Economics (AREA)
  • Quality & Reliability (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Secondary Cells (AREA)
  • Electric Propulsion And Braking For Vehicles (AREA)

Abstract

本发明提供了一种电动汽车的电池绝缘失效预测方法，包括：机器学习平台对从大数据平台获取的车辆历史数据进行合并和分割处理，得到至少两类电池绝缘失效影响数据；机器学习平台采用电池绝缘失效影响数据对预设的线性分类器进行训练，得到电池绝缘失效预测模型；业务运营平台根据车端的当前电池绝缘失效影响数据，应用机器学习平台发送的电池绝缘失效预测模型进行离线电池绝缘失效风险预测；或者，车端根据其当前电池绝缘失效影响数据，应用电池绝缘失效预测模型进行在线电池绝缘失效风险预测；本发明可以实现车端的实时在线电池绝缘失效风险预测和业务运营平台的离线电池绝缘失效风险预测，提升电池绝缘失效风险预测的可靠性。

Figure 202111263384

The invention provides a battery insulation failure prediction method for electric vehicles, which includes: a machine learning platform merges and divides historical vehicle data obtained from a big data platform to obtain at least two types of battery insulation failure impact data; the machine learning platform adopts The battery insulation failure impact data is used to train a preset linear classifier to obtain a battery insulation failure prediction model; the business operation platform applies the battery insulation failure prediction model sent by the machine learning platform to offline battery insulation according to the current battery insulation failure impact data at the vehicle end. Failure risk prediction; alternatively, the vehicle end uses a battery insulation failure prediction model to perform online battery insulation failure risk prediction according to its current battery insulation failure impact data; the present invention can realize real-time online battery insulation failure risk prediction at the vehicle end and offline business operation platform. Battery insulation failure risk prediction improves the reliability of battery insulation failure risk prediction.

Figure 202111263384

Description

Battery insulation failure prediction method for electric automobile
Technical Field
The invention relates to the technical field of battery failure prediction of electric automobiles, in particular to a battery insulation failure prediction method of an electric automobile.
Background
With the rapid development of electric automobiles in recent years, the popularity of electric automobiles has become higher and higher. Safety accidents of the electric vehicle also occur frequently, wherein safety accidents caused by the power battery are increased gradually, so that safety problems of the power battery are paid more and more attention. At present, the existing scheme for early warning of the insulation failure of the battery generally adopts the analysis of battery data cores of the power battery, such as the battery temperature, the battery voltage, the insulation internal resistance of the battery, and the like, does not consider the influence of the driving habit of a driver, the vehicle working condition, the external environment and the like on the insulation failure of the power battery, and leads the reliability of the insulation failure prediction result of the power battery to be poor.
Disclosure of Invention
In view of the above problems, an object of the present invention is to provide a method for predicting insulation failure of a battery of an electric vehicle, which can realize real-time online prediction of insulation failure risk of the battery at a vehicle end and offline prediction of insulation failure risk of the battery at a service operation platform, and improve reliability of prediction of insulation failure risk of the battery.
The embodiment of the invention provides a method for predicting the insulation failure of a battery of an electric automobile, which comprises the following steps:
the machine learning platform is used for merging and dividing the vehicle historical data acquired from the big data platform to obtain at least two types of battery insulation failure influence data;
the machine learning platform trains a preset linear classifier by adopting the battery insulation failure influence data to obtain a battery insulation failure prediction model;
the service operation platform applies the battery insulation failure prediction model sent by the machine learning platform to carry out offline battery insulation failure risk prediction according to the current battery insulation failure influence data of the vehicle end; or the vehicle end applies the battery insulation failure prediction model to perform online battery insulation failure risk prediction according to the current battery insulation failure influence data.
As an improvement of the above scheme, the machine learning platform trains a preset linear classifier by using the battery insulation failure influence data to obtain a battery insulation failure prediction model, including:
the machine learning platform slices the battery insulation failure influence data to obtain a plurality of data slices; wherein each data slice is equal in length;
and the machine learning platform trains a preset linear classifier according to the data slice to obtain a battery insulation failure prediction model.
As an improvement of the above scheme, the machine learning platform performs merging and segmentation processing on vehicle history data acquired from a big data platform to obtain at least two types of battery insulation failure influence data, including:
the machine learning platform merges the vehicle historical data according to the vehicle to which the vehicle historical data belongs and the timestamp;
and the machine learning platform divides the merged vehicle historical data according to the vehicle state to obtain at least two types of battery insulation failure influence data.
As a modification of the above, the vehicle state includes a high-voltage state on the battery, a state of charge;
then, the machine learning platform divides the merged vehicle history data according to the vehicle state to obtain at least two types of battery insulation failure influence data, including:
the machine learning platform segments the merged vehicle history data into battery insulation failure impact data corresponding to a high voltage condition on the battery and battery insulation failure impact data corresponding to the state of charge.
As an improvement of the above, the method further comprises:
before slicing processing, the machine learning platform performs dimension reduction processing on the battery insulation failure influence data;
after the slicing process, the machine learning platform normalizes and normalizes the data slices.
As an improvement of the above, the method further comprises:
and the machine learning platform performs data cleaning and/or data missing filling processing on the reduced battery insulation failure influence data.
As an improvement of the above scheme, the performing, by the machine learning platform, the dimension reduction processing on the battery insulation failure influence data includes:
the machine learning platform analyzes the Pearson correlation coefficient of the battery insulation failure influence data, screens out the battery insulation failure influence data with the Pearson coefficient smaller than a first set value, and takes the battery insulation failure influence data as the reduced-dimension battery insulation failure influence data;
or the machine learning platform carries out PCA dimension reduction on the battery insulation failure influence data to obtain the dimension-reduced battery insulation failure influence data.
As an improvement of the above scheme, the data missing filling processing specifically includes:
when one of the battery insulation failure influence data belongs to event type data, missing filling is carried out on the data by adopting data corresponding to a timestamp;
when one of the battery insulation failure influence data belongs to continuous periodic data, performing missing filling on the data in an average filling mode;
and when one data in the battery insulation failure influence data belongs to discrete periodic data, performing missing filling on the data by adopting the data corresponding to the last timestamp.
As an improvement of the scheme, the battery insulation failure influence data comprise driving behavior data, whole vehicle data, vehicle working condition data, battery data and environment data.
As an improvement of the above scheme, the data washing specifically includes:
and performing data cleaning on the reduced battery insulation failure influence data by adopting a normally distributed 3 sigma principle, and removing invalid data and repeated data in the battery insulation failure influence data.
As an improvement of the above, the method comprises:
the service operation platform carries out monitoring and early warning according to the battery insulation failure risk predicted by the service operation platform or the battery insulation failure risk predicted by the vehicle end;
or the user side or the vehicle-mounted system carries out early warning according to the battery insulation failure risk obtained by the vehicle-mounted prediction.
Compared with the prior art, the embodiment of the invention has the beneficial effects that: vehicle historical data of a pre-existing big data platform are obtained; merging and dividing historical data of the vehicle to obtain at least two types of battery insulation failure influence data; preprocessing the battery insulation failure influence data, and training a preset linear classifier by using the battery insulation failure influence data to obtain a battery insulation failure prediction model; the vehicle end applies the battery insulation failure prediction model to carry out online battery insulation failure risk prediction according to the current battery insulation failure influence data; meanwhile, the service operation platform applies a battery insulation failure prediction model sent by the machine learning platform to perform offline battery insulation failure risk prediction according to current battery insulation failure influence data of a vehicle end; the invention adopts a big data platform to pre-store a mass of human-vehicle-environment vehicle historical data for machine learning, obtains a battery insulation failure prediction model and transmits the battery insulation failure prediction model to the vehicle end and the service operation platform, thereby realizing the real-time online battery insulation failure risk prediction of the vehicle end and the offline battery insulation failure risk prediction of the service operation platform, and improving the reliability of the battery insulation failure risk prediction.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a method for predicting insulation failure of a battery of an electric vehicle according to an embodiment of the present invention;
fig. 2 is a schematic block diagram of battery insulation failure prediction provided by an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, the present invention provides a method for predicting an insulation failure of a battery of an electric vehicle, including:
s1: the machine learning platform is used for merging and dividing the vehicle historical data acquired from the big data platform to obtain at least two types of battery insulation failure influence data;
further, the battery insulation failure influence data comprise driving behavior data, whole vehicle data, vehicle working condition data, battery data and environment data.
Illustratively, each control domain of the vehicle, such as a VDCM domain, a BDCM domain, and an IDVM domain, uploads its own current domain data, such as VDCM domain data, BDCM domain data, and IDVM domain data, to the big data platform in real time through the vehicle-mounted gateway T-BOX, that is, the vehicle history data stored in the big data platform includes data uploaded by each control domain of the vehicle end. The big data platform analyzes and primarily screens real-time domain data uploaded by a vehicle controller, divides the real-time domain data into 5 types of data, and respectively stores the 5 types of data in different data sources, wherein the data comprise finished vehicle data, vehicle working condition data, battery data, driving behavior data and environment data; the vehicle working condition data comprises tire pressure, tire temperature, temperature in the vehicle, air conditioner power, six doors, two covers, lamps, motor rotating speed, speed and the like; the battery data comprises current, voltage, battery temperature, SOC, fast charge, slow charge, battery characteristics in driving, battery characteristics in charging and the like; the driving behavior data comprises vehicle speed, wheel speed, air conditioner control, brake plate force, safety belt control, lamp control, parking, gear and the like; the environment data comprises weather information such as rainfall, weather, temperature and humidity, road condition information such as mountainous areas, cities, high speed and speed limit, and can also comprise charging pile information in charging. The preliminary screening of the real-time domain data can be carried out according to the influence degree/correlation degree of the data on the driving mileage of the vehicle, so that data irrelevant to driving mileage estimation is eliminated, and whether the data are relevant to driving mileage estimation or not can be judged according to expert experience. The big data platform stores signal data of different types of more than 200 of the same type and different vehicles, and improves massive data basic support for subsequent mileage prediction. Wherein each vehicle history data carries a timestamp and vehicle status information.
It should be noted that the processing of the uploaded vehicle history data by the big data platform belongs to the prior art, and will not be described in detail herein, and different vehicle history data can be stored to different data sources through the big data platform, for example, according to static data, dynamic data, structured data, semi-structured data, and unstructured data.
Because the endurance belongs to time sequence data, the machine learning platform extracts the vehicle historical data from the big data platform according to the sequence of the timestamps and constructs a model.
Further, screening the vehicle historical data;
for example, after the data magnitude is determined, vehicle history data of N vehicles in a set time period is screened out, and for example, vehicle history data of 100 vehicles for 90 days is extracted from the big data platform according to the sequence of timestamps to serve as an initial data set.
Further, S1: the machine learning platform merges and cuts apart the vehicle historical data that obtains from big data platform, obtains two at least types of battery insulation failure influence data, includes:
the machine learning platform merges the vehicle historical data according to the vehicle to which the vehicle historical data belongs and the timestamp;
examples of the merged vehicle history data are given as described in the following table.
Vehicle frame number Time stamp Data 1 Data 2 Data 3 Data 4 ... Data M Remarks for note
Vehicle 1 1633609494000 3 0 444.5 72.1 ... 2 Dormancy
Vehicle 1 1633609494000 3 0 444.5 72.1 ... 2 Quick charger
Vehicle 2 1633609495000 3 0 369 72.1 ... 0 Slow charging
Vehicle 2 1633609496000 2 0 78.75 72.1 ... 0 Upper high pressure
... ... ... ... ... ... ... ... ...
Vehicle N 1633609497000 2 0 14 72.1 ... 0 Upper high pressure
And the machine learning platform divides the merged vehicle historical data according to the vehicle state to obtain at least two types of battery insulation failure influence data.
The vehicle state includes a high voltage state on the battery (i.e., a driving state), a charging state (including a rapid charging, a slow charging), a resting state, and the like. Because the factors influencing the endurance are different when the vehicle is in different states, when the battery insulation failure influence data are segmented, appropriate data need to be segmented from the vehicle historical data according to the vehicle states corresponding to different timestamps, for example, in the charging process, data such as speed and acceleration cannot be used as the battery insulation failure influence data, and in the driving process, the charging pile and the discharging gun cannot be used as the battery insulation failure influence data. Through data merging and dividing processing, the historical data of the vehicle can be divided into different types of battery insulation failure influence data, so that the prediction of the endurance mileage is more consistent with the current state of the vehicle.
S2: the machine learning platform trains a preset linear classifier by adopting the battery insulation failure influence data to obtain a battery insulation failure prediction model;
s3: the service operation platform applies the battery insulation failure prediction model sent by the machine learning platform to carry out offline battery insulation failure risk prediction according to the current battery insulation failure influence data of the vehicle end; or the vehicle end applies the battery insulation failure prediction model to perform online battery insulation failure risk prediction according to the current battery insulation failure influence data.
The embodiment of the invention adopts a big data platform to pre-store a mass of human-vehicle-environment vehicle historical data for machine learning, obtains a battery insulation failure prediction model and sends the battery insulation failure prediction model to a vehicle end, so that the vehicle end carries out online edge calculation through the battery insulation failure prediction model to obtain the battery insulation failure risk corresponding to the current battery insulation failure influence data, and meanwhile, the service operation platform carries out offline battery insulation failure risk prediction by using the battery insulation failure prediction model sent by the machine learning platform according to the current battery insulation failure influence data of the vehicle end, thereby realizing the real-time online battery insulation failure risk prediction of the vehicle end and the offline battery insulation failure risk prediction of the service operation platform and improving the reliability of the battery insulation failure risk prediction.
In an alternative embodiment, the vehicle state includes a high voltage on battery state, a state of charge;
then, the dividing the merged vehicle history data according to the vehicle state to obtain at least two types of battery insulation failure influence data includes:
the merged vehicle history data is partitioned into battery insulation failure impact data corresponding to a high voltage condition on the battery and battery insulation failure impact data corresponding to the state of charge.
In an optional embodiment, the training of the preset linear classifier by the machine learning platform using the battery insulation failure influence data to obtain a battery insulation failure prediction model includes:
the machine learning platform slices the battery insulation failure influence data to obtain a plurality of data slices; wherein each data slice is equal in length;
and the machine learning platform trains a preset linear classifier according to the data slice to obtain a battery insulation failure prediction model.
In the embodiment of the invention, the data is sliced by taking the set time length as an interval. For example, assuming that there are 100 pieces of data, 1 piece of data per second, slicing is performed in 30s as an interval, so that 3 ten thousand data slices can be obtained, slicing is performed in 300s as an interval, so that 3 ten thousand data slices, 3000 data slices can be obtained, and each data slice is used as a sample and input to a preset linear classifier, such as an SVM classifier, for training.
Further, in order to improve the validity of the data, the intervals need to be filtered, specifically, all the intervals are traversed, and when the data amount of the next interval is smaller than the set number, the interval is rejected; and the numerical value of the set quantity is equal to the numerical value of the set duration corresponding to the interval. I.e. the data slices after interval filtering are of equal length.
In an optional embodiment, the method further comprises:
before slicing processing, the machine learning platform performs dimension reduction processing on the battery insulation failure influence data;
illustratively, the battery insulation failure influence data can be subjected to dimension reduction processing in the modes of average number, median, standard deviation, data visualization, histogram, Pearson correlation coefficient, PAC dimension reduction and the like, and data with poor relevance to endurance are eliminated.
Further, the machine learning platform performs dimension reduction processing on the battery insulation failure influence data, and the dimension reduction processing comprises:
the machine learning platform analyzes the Pearson correlation coefficient of the battery insulation failure influence data, screens out the battery insulation failure influence data with the Pearson coefficient smaller than a first set value, and takes the battery insulation failure influence data as the reduced-dimension battery insulation failure influence data;
or the machine learning platform carries out PCA dimension reduction on the battery insulation failure influence data to obtain the dimension-reduced battery insulation failure influence data.
It should be noted that the pearson correlation coefficient is defined as the quotient of the covariance and the standard deviation between two data, and a linear correlation coefficient is expressed; wherein, the data are linear relations and are continuous data; the population of data is a normal distribution, or a unimodal distribution that is nearly normal; the observations of the data are paired, and each pair of observations is independent of each other.
Calculating a probability value of no correlation between the two data, namely a Pearson coefficient P value, and then calculating a coefficient, namely an R value;
if the Pearson coefficient P is small, the R value is valid, and the table indicates that the two data have correlation, the two data are retained. Illustratively, setting the first set value to 0.05 and the second set value to 0.5 may screen out data with a pearson coefficient P value >0.5 and a P value <0.05, implement dimensionality reduction of the data, obtain data with a P value <0.05, and reject weakly related or unrelated data.
In other embodiments, the collected data may be analyzed according to the unit of average, median, and mode, such as the temperature in the vehicle, the average temperature for the tire pressure, the mode for the resistance, and the like, and the median for the speed, and the like. For example, the analysis of the remote control T-Box power data shows that whether the average value, the mode or the median is not consistent with the actual operation, the data has quality problems, the data is regarded as the characteristic is not qualified, and the data is rejected.
The PAC dimension reduction principle is as follows:
and m n-dimensional battery insulation failure influence data are set.
Forming n rows and m columns of matrix X by the original battery insulation failure influence data according to columns;
zero-averaging each row of the X, namely subtracting the mean value of the row, and then solving the covariance matrix to calculate the eigenvalue of the covariance matrix and the corresponding eigenvector;
and arranging the eigenvectors into a matrix from top to bottom according to the size of the corresponding eigenvalue, and taking the first k rows to form a matrix P, namely the battery insulation failure influence data after the dimension is reduced to k dimensions.
After the slicing process, the machine learning platform normalizes and normalizes the data slices.
In the embodiment of the invention, the problem of data attribute dimensionless can be solved by carrying out standardization and normalization processing on the battery insulation failure influence data. In particular, by the formula
Figure BDA0003326271580000101
Performing a normalization process by a formula
Figure BDA0003326271580000102
And (6) carrying out normalization processing. Wherein, mu represents the average value of the battery insulation failure influence data, sigma represents the standard deviation of the battery insulation failure influence data, X represents a battery insulation failure influence data, X representsminMinimum value, X, representing battery insulation failure influence datamaxRepresents the maximum value of the battery insulation failure influence data.
In an optional embodiment, the method further comprises:
and the machine learning platform performs data cleaning and/or data missing filling processing on the reduced battery insulation failure influence data.
For example, the data cleaning may be performed by using a normal distribution 3 σ principle, a value range, NA cleaning, a deduplication rule, and the like, and in the embodiment of the present invention, the data cleaning is performed by preferably using the normal distribution 3 σ principle.
Further, the data cleaning specifically includes:
and performing data cleaning on the reduced battery insulation failure influence data by adopting a normally distributed 3 sigma principle, and removing invalid data and repeated data in the battery insulation failure influence data.
For example, based on the 3 σ principle of normal distribution, a value with a deviation exceeding 3 times of the standard deviation σ from the average value μmay be screened out, and if the data follows normal distribution, the probability of occurrence of values other than 3 σ from the average value is P (| x- μ | >3 σ) < ═ 0.003, belonging to a small probability event, the data is rejected.
Filling the data missing, and specifically processing;
when one of the battery insulation failure influence data belongs to event type data, missing filling is carried out on the data by adopting data corresponding to a timestamp;
when one of the battery insulation failure influence data belongs to continuous periodic data, performing missing filling on the data in an average filling mode;
and when one data in the battery insulation failure influence data belongs to discrete periodic data, performing missing filling on the data by adopting the data corresponding to the last timestamp.
Because the data is collected according to the appointed frequency in the time stamp sequence, the data is lost and the wrong data is collected due to the influence of the network, the sensor, the performance and the like in the collection process, and the invalid data is generated, the corresponding event type missing data can be filled according to the data corresponding to the last time stamp, the continuous periodic missing data can be filled according to the average value, the discrete periodic data is filled in the missing manner according to the data of the last time stamp, and the continuous effective data sequence is ensured.
In an alternative embodiment, the linear classifier is a SVM classifier;
then, the machine learning platform trains a preset linear classifier according to the data slice to obtain a battery insulation failure prediction model, including:
dividing the data slice into a training sample set, a verification sample set and a test sample set;
sequentially taking the data slices in the training sample set as input feature data of the SVM classifier, taking the battery insulation failure risk corresponding to the data slices as the output feature of the SVM classifier, and constructing a battery insulation failure prediction model; wherein the risk of battery insulation failure comprises: failure and normal;
performing model verification on the battery insulation failure prediction model by adopting the verification sample set;
and performing model test on the battery insulation failure prediction model by adopting the test sample set.
In the embodiment of the present invention, the SVM classifier constructs a linear function by using a ploy kernel (i.e., a polynomial kernel).
In the embodiment of the invention, the data is divided into a training set, a verification set and a test set according to a set proportion, for example, a proportion of 60%, 20% and 20%, and model training, verification and testing are sequentially performed to ensure that the prediction accuracy of the constructed battery insulation failure prediction model meets the expectation.
After a battery insulation failure prediction model is constructed, the battery insulation failure prediction model is transmitted to a domain controller at the vehicle end, so that the domain controller at the vehicle end can predict the battery insulation failure through the battery insulation failure prediction model and the collected battery insulation failure influence data at the current moment. Because the construction and the subsequent iteration updating of the battery insulation failure influence data do not need to be executed at the vehicle end, the vehicle end only needs to carry out edge calculation, the calculated amount of the vehicle end can be effectively reduced, and the efficiency of predicting the endurance mileage of the vehicle end is improved.
In an alternative embodiment, the method comprises:
the service operation platform carries out monitoring and early warning according to the battery insulation failure risk predicted by the service operation platform or the battery insulation failure risk predicted by the vehicle end;
or the user side or the vehicle-mounted system carries out early warning according to the battery insulation failure risk obtained by the vehicle-mounted prediction.
As shown in fig. 2, after the machine learning platform constructs a battery insulation failure prediction model, the machine learning platform sends the battery insulation failure prediction model to the service operation platform, so that the service operation platform applies the battery insulation failure prediction model to perform offline battery insulation failure risk prediction, thereby realizing batch offline monitoring and early warning of monitoring vehicle ends; meanwhile, the service operation platform sends the battery insulation failure prediction model to the vehicle end through the TSP cloud platform and the T-BOX, so that the vehicle end applies the battery insulation failure prediction model to perform online battery insulation failure risk prediction. After the vehicle end predicts the battery insulation failure risk, uploading the battery insulation failure risk to a vehicle machine system of the vehicle through the T-BOX to perform vehicle machine early warning, and simultaneously uploading the battery insulation failure risk to a user side APP through the T-BOX and a TSP cloud platform to perform early warning; furthermore, the vehicle end sends the battery insulation failure risk to the service operation platform through the T-BOX and the TSP cloud platform for monitoring and early warning, and meanwhile, the service operation platform monitors and early warns the battery insulation failure risk obtained by self prediction, so that synchronous early warning of the battery insulation failure risk of the vehicle end and the service operation platform can be realized.
Compared with the prior art, the embodiment of the invention has the beneficial effects that: vehicle historical data of a pre-existing big data platform are obtained; merging and dividing historical data of the vehicle to obtain at least two types of battery insulation failure influence data; training a preset linear classifier by using the battery insulation failure influence data to obtain a battery insulation failure prediction model; the vehicle end applies the battery insulation failure prediction model to carry out online battery insulation failure risk prediction according to the current battery insulation failure influence data; meanwhile, the service operation platform applies a battery insulation failure prediction model sent by the machine learning platform to perform offline battery insulation failure risk prediction according to current battery insulation failure influence data of a vehicle end; the invention adopts a big data platform to pre-store a mass of human-vehicle-environment vehicle historical data for machine learning, obtains a battery insulation failure prediction model and transmits the battery insulation failure prediction model to the vehicle end and the service operation platform, thereby realizing the real-time online battery insulation failure risk prediction of the vehicle end and the offline battery insulation failure risk prediction of the service operation platform, and improving the reliability of the battery insulation failure risk prediction.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention.

Claims (11)

1.一种电动汽车的电池绝缘失效预测方法，其特征在于，1. a battery insulation failure prediction method of electric vehicle, is characterized in that, 机器学习平台对从大数据平台获取的车辆历史数据进行合并和分割处理，得到至少两类电池绝缘失效影响数据；The machine learning platform merges and divides the historical vehicle data obtained from the big data platform to obtain at least two types of battery insulation failure impact data; 所述机器学习平台采用所述电池绝缘失效影响数据对预设的线性分类器进行训练，得到电池绝缘失效预测模型；The machine learning platform uses the battery insulation failure impact data to train a preset linear classifier to obtain a battery insulation failure prediction model; 所述业务运营平台根据车端的当前电池绝缘失效影响数据，应用所述机器学习平台发送的电池绝缘失效预测模型进行离线电池绝缘失效风险预测；或者，车端根据其当前电池绝缘失效影响数据，应用所述电池绝缘失效预测模型进行在线电池绝缘失效风险预测。The business operation platform uses the battery insulation failure prediction model sent by the machine learning platform to perform offline battery insulation failure risk prediction according to the current battery insulation failure impact data on the vehicle end; The battery insulation failure prediction model performs online battery insulation failure risk prediction. 2.如权利要求1所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述机器学习平台采用所述电池绝缘失效影响数据对预设的线性分类器进行训练，得到电池绝缘失效预测模型，包括：2 . The method for predicting battery insulation failure of an electric vehicle according to claim 1 , wherein the machine learning platform uses the battery insulation failure impact data to train a preset linear classifier to obtain a battery insulation failure prediction. 3 . models, including: 所述机器学习平台对所述电池绝缘失效影响数据进行切片处理，得到若干数据切片；其中，每个数据切片的长度相等；The machine learning platform performs slice processing on the battery insulation failure impact data to obtain several data slices; wherein the lengths of each data slice are equal; 所述机器学习平台根据所述数据切片对预设的线性分类器进行训练，得到电池绝缘失效预测模型。The machine learning platform trains a preset linear classifier according to the data slice to obtain a battery insulation failure prediction model. 3.如权利要求1所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述机器学习平台对从大数据平台获取的车辆历史数据进行合并和分割处理，得到至少两类电池绝缘失效影响数据，包括：3. The method for predicting battery insulation failure of an electric vehicle according to claim 1, wherein the machine learning platform merges and divides the historical vehicle data obtained from the big data platform to obtain at least two types of battery insulation failures Impact data, including: 所述机器学习平台将所述车辆历史数据按照其所属的车辆及时间戳进行合并；The machine learning platform merges the vehicle historical data according to the vehicle to which it belongs and the timestamp; 所述机器学习平台将合并后的车辆历史数据按照车辆状态进行分割，得到至少两类电池绝缘失效影响数据。The machine learning platform divides the merged historical data of the vehicle according to the state of the vehicle, and obtains at least two types of battery insulation failure impact data. 4.如权利要求3所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述车辆状态包括电池上高压状态、充电状态；4. The method for predicting battery insulation failure of an electric vehicle according to claim 3, wherein the vehicle state comprises a high voltage state and a charging state on the battery; 则，所述机器学习平台将合并后的车辆历史数据按照车辆状态进行分割，得到至少两类电池绝缘失效影响数据，包括：Then, the machine learning platform divides the merged vehicle historical data according to the vehicle state, and obtains at least two types of battery insulation failure impact data, including: 所述机器学习平台将合并后的车辆历史数据分割成对应所述电池上高压状态的电池绝缘失效影响数据和对应所述充电状态的电池绝缘失效影响数据。The machine learning platform divides the combined vehicle historical data into battery insulation failure impact data corresponding to the high voltage state on the battery and battery insulation failure impact data corresponding to the charging state. 5.如权利要求2所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述方法还包括：5. The method for predicting battery insulation failure of an electric vehicle according to claim 2, wherein the method further comprises: 在切片处理之前，所述机器学习平台对所述电池绝缘失效影响数据进行降维处理；Before slicing processing, the machine learning platform performs dimensionality reduction processing on the battery insulation failure impact data; 在切片处理之后，所述机器学习平台对所述数据切片进行标准化和归一化处理。After slice processing, the machine learning platform normalizes and normalizes the data slices. 6.如权利要求5所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述方法还包括：6. The method for predicting battery insulation failure of an electric vehicle according to claim 5, wherein the method further comprises: 所述机器学习平台对降维后的电池绝缘失效影响数据进行数据清洗和/或数据缺失填充处理。The machine learning platform performs data cleaning and/or data missing filling processing on the battery insulation failure impact data after dimension reduction. 7.如权利要求5所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述机器学习平台对所述电池绝缘失效影响数据进行降维处理，包括：7. The method for predicting battery insulation failure of an electric vehicle according to claim 5, wherein the machine learning platform performs dimensionality reduction processing on the battery insulation failure impact data, comprising: 所述机器学习平台对所述电池绝缘失效影响数据进行皮尔逊相关系数分析，筛选出皮尔逊系数小于第一设定值的电池绝缘失效影响数据，作为降维后的电池绝缘失效影响数据；The machine learning platform performs a Pearson correlation coefficient analysis on the battery insulation failure impact data, and screens out battery insulation failure impact data whose Pearson coefficient is less than a first set value, as the battery insulation failure impact data after dimensionality reduction; 或者，所述机器学习平台对所述电池绝缘失效影响数据进行PCA降维，得到降维后的电池绝缘失效影响数据。Alternatively, the machine learning platform performs PCA dimension reduction on the battery insulation failure impact data to obtain the battery insulation failure impact data after dimension reduction. 8.如权利要求6所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述数据缺失填充处理，具体包括：8. The method for predicting battery insulation failure of an electric vehicle according to claim 6, wherein the data missing filling process specifically comprises: 当所述电池绝缘失效影响数据中的一个数据属于事件型数据时，采用对应上一时间戳的数据对所述数据进行缺失填充；When one data in the battery insulation failure impact data belongs to event data, the data corresponding to the last timestamp is used to fill in the missing data; 当所述电池绝缘失效影响数据中的一个数据属于连续周期型数据时，采用平均值填充方式对所述数据进行缺失填充；When one data in the battery insulation failure impact data belongs to continuous periodic data, the data is filled with missing data by means of average value filling; 当所述电池绝缘失效影响数据中的一个数据属于离散周期型数据时，采用对应上一时间戳的数据对所述数据进行缺失填充。When one data in the battery insulation failure impact data belongs to discrete periodic data, the data corresponding to the last time stamp is used to fill in the missing data. 9.如权利要求1所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述电池绝缘失效影响数据包括驾驶行为数据、整车数据、车辆工况数据、电池数据以及环境数据。9 . The method for predicting battery insulation failure of an electric vehicle according to claim 1 , wherein the battery insulation failure impact data includes driving behavior data, vehicle data, vehicle operating condition data, battery data, and environmental data. 10 . 10.如权利要求6所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述数据清洗，具体包括：10. The method for predicting battery insulation failure of an electric vehicle according to claim 6, wherein the data cleaning specifically comprises: 采用正态分布的3σ原则对降维后的电池绝缘失效影响数据进行数据清洗，剔除所述电池绝缘失效影响数据中无效数据和重复数据。The 3σ principle of normal distribution is used to clean the battery insulation failure impact data after dimensionality reduction, and the invalid data and duplicate data in the battery insulation failure impact data are eliminated. 11.如权利要求1所述的电动汽车的电池绝缘失效预测方法，其特征在于，所述方法包括：11. The method for predicting battery insulation failure of an electric vehicle according to claim 1, wherein the method comprises: 所述业务运营平台根据其自身预测得到的电池绝缘失效风险或者所述车端预测得到的电池绝缘失效风险进行监控预警；The business operation platform performs monitoring and early warning according to the battery insulation failure risk predicted by itself or the battery insulation failure risk predicted by the vehicle end; 或者，用户端或车机系统根据所述车端预测得到的电池绝缘失效风险进行预警。Alternatively, the user terminal or the on-board system may issue an early warning according to the risk of battery insulation failure predicted by the on-board terminal.
CN202111263384.2A 2021-10-28 2021-10-28 A battery insulation failure prediction method for electric vehicles Pending CN114118217A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111263384.2A CN114118217A (en) 2021-10-28 2021-10-28 A battery insulation failure prediction method for electric vehicles

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111263384.2A CN114118217A (en) 2021-10-28 2021-10-28 A battery insulation failure prediction method for electric vehicles

Publications (1)

Publication Number Publication Date
CN114118217A true CN114118217A (en) 2022-03-01

Family

ID=80377525

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111263384.2A Pending CN114118217A (en) 2021-10-28 2021-10-28 A battery insulation failure prediction method for electric vehicles

Country Status (1)

Country Link
CN (1) CN114118217A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115884233A (en) * 2022-10-26 2023-03-31 中国航空工业集团公司西安航空计算技术研究所 A Highly Reliable Airborne Tire Pressure Monitoring Method
WO2023164965A1 (en) * 2022-03-04 2023-09-07 福建时代星云科技有限公司 Direct-current insulation monitoring method and terminal for energy storage system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110457833A (en) * 2019-08-15 2019-11-15 深圳普瑞赛思检测技术有限公司 A kind of data construction method for cell safety and performance evaluation
CN111277221A (en) * 2020-02-13 2020-06-12 中国联合网络通信集团有限公司 A photovoltaic fault diagnosis method and device
CN112380630A (en) * 2020-11-27 2021-02-19 重庆长安新能源汽车科技有限公司 New energy automobile battery thermal runaway risk prediction method based on Internet of vehicles data
CN112950811A (en) * 2021-02-02 2021-06-11 武汉理工大学 New energy automobile region operation risk assessment and early warning system integrating whole automobile safety

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110457833A (en) * 2019-08-15 2019-11-15 深圳普瑞赛思检测技术有限公司 A kind of data construction method for cell safety and performance evaluation
CN111277221A (en) * 2020-02-13 2020-06-12 中国联合网络通信集团有限公司 A photovoltaic fault diagnosis method and device
CN112380630A (en) * 2020-11-27 2021-02-19 重庆长安新能源汽车科技有限公司 New energy automobile battery thermal runaway risk prediction method based on Internet of vehicles data
CN112950811A (en) * 2021-02-02 2021-06-11 武汉理工大学 New energy automobile region operation risk assessment and early warning system integrating whole automobile safety

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023164965A1 (en) * 2022-03-04 2023-09-07 福建时代星云科技有限公司 Direct-current insulation monitoring method and terminal for energy storage system
CN115884233A (en) * 2022-10-26 2023-03-31 中国航空工业集团公司西安航空计算技术研究所 A Highly Reliable Airborne Tire Pressure Monitoring Method

Similar Documents

Publication Publication Date Title
CN112487617B (en) Collision model-based risk prevention method, device, equipment and storage medium
CN108303264B (en) Cloud-based vehicle fault diagnosis method, device and system
KR101617349B1 (en) Diagnostic system and method for the analysis of driving behavior
US11599563B2 (en) Programmatically identifying a personality of an autonomous vehicle
CN105160883A (en) Energy-saving driving behavior analysis method based on big data
CN115423035A (en) User portrait generation method based on feature variable scoring, equipment, cars and storage media
CN114118217A (en) A battery insulation failure prediction method for electric vehicles
CN112990545B (en) A method for predicting traffic safety status in highway weaving area
CN113485301B (en) A quantitative evaluation method for the expected functional safety performance of autonomous vehicle control systems
CN110012069A (en) A kind of vehicle data processing unit, method and vehicle monitoring platform
CN112766578A (en) Vehicle use identification method and system based on vehicle network and storage medium
CN114971156A (en) Driving style analysis method based on big data dangerous scene
CN114091735A (en) Electric vehicle endurance mileage prediction method
CN118691321B (en) Online car market automobile substation data management and control platform based on edge computing
CN114771502A (en) Energy consumption optimization method, device, equipment, storage medium and system
CN114444872A (en) Method and system for analyzing driving behavior of driver by adopting artificial intelligence algorithm
CN102945601B (en) Traffic incident duration forecasting method based on partial least squares method
GB2598785A (en) Monitoring system
WO2024168454A1 (en) Vehicle chassis health management method, apparatus and system
CN117743802A (en) Vehicle-cloud-coordination-based all-weather and multi-region power battery pack SOH prediction method and system
CN116794544A (en) Method for detecting life of power battery, computer-readable storage medium, and vehicle
CN116552249A (en) Vehicle remaining mileage prediction method, device, equipment and storage medium
CN115545102A (en) Vehicle energy consumption monitoring method based on random forest
CN118656594B (en) Cloud-based safety monitoring system and monitoring method based on single-vehicle-multi-vehicle fusion
CN114919413B (en) A method and system for diagnosing abnormality of an electric vehicle battery

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination