CN116381500A - Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis - Google Patents
Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis Download PDFInfo
- Publication number
- CN116381500A CN116381500A CN202310160821.0A CN202310160821A CN116381500A CN 116381500 A CN116381500 A CN 116381500A CN 202310160821 A CN202310160821 A CN 202310160821A CN 116381500 A CN116381500 A CN 116381500A
- Authority
- CN
- China
- Prior art keywords
- parameter
- vehicle
- data
- parameters
- power battery
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 100
- 238000011156 evaluation Methods 0.000 title claims abstract description 25
- 238000007405 data analysis Methods 0.000 title claims abstract description 18
- 238000005070 sampling Methods 0.000 claims abstract description 99
- 238000003062 neural network model Methods 0.000 claims abstract description 62
- 230000008569 process Effects 0.000 claims abstract description 21
- 238000000611 regression analysis Methods 0.000 claims abstract description 8
- 230000001419 dependent effect Effects 0.000 claims abstract description 6
- 238000012163 sequencing technique Methods 0.000 claims description 52
- 230000005855 radiation Effects 0.000 claims description 30
- 238000010219 correlation analysis Methods 0.000 claims description 27
- 230000001133 acceleration Effects 0.000 claims description 23
- 238000013527 convolutional neural network Methods 0.000 claims description 22
- 238000004891 communication Methods 0.000 claims description 15
- 230000007613 environmental effect Effects 0.000 claims description 12
- 238000013178 mathematical model Methods 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 11
- 125000004122 cyclic group Chemical group 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 5
- 101000872083 Danio rerio Delta-like protein C Proteins 0.000 claims description 4
- 238000005286 illumination Methods 0.000 claims description 4
- 238000010295 mobile communication Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 abstract description 10
- 238000013528 artificial neural network Methods 0.000 abstract description 2
- 238000010276 construction Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000002238 attenuated effect Effects 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 230000002441 reversible effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000000275 quality assurance Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention discloses a method for constructing an evaluation index of a power battery of a pure electric vehicle based on big data analysis, which comprises the following steps: A1. marking the vehicle, A2, P 1 In the running and using process of the vehicle, A3 and P are as follows 2 ……P n Repeating steps A2 and A5 and eliminating No. 1 DB for new vehicles ND Parameters are distributed according to 1% of front data and 99% of rear data after normal distribution, A6, a calibration sample is taken as a target value, A7, and a neural network is usedModel MM 1 ……MM n And (3) taking the calculated value at each sampling point as a dependent variable, A8, using a regression analysis method, A9, and additionally extracting one or more regression equations with the lowest comprehensive error rate from a plurality of regression equations as a continuous parameter prediction model. Compared with the traditional neural network model calculation method, the method has the advantages of greatly reduced calculation amount, better accuracy and adaptability compared with a single correlation regression prediction method, and wide practical application value.
Description
Technical Field
The invention relates to the technical field of power battery evaluation, in particular to a method for constructing a power battery evaluation index of a pure electric vehicle based on big data analysis.
Background
At present, the parameters of the testability index in the production process of the new battery of the power battery of the pure electric automobile are very many, wherein the parameters of the national standard mandatory requirements and the parameters of the self-assessment use of various factories exist;
but for the consumer, where the most important parameter indicator is battery capacity decay rate, FIG. 1 is Tesla Model S/Model X20 ten thousand miles (about 32 ten thousand kilometers) battery capacity retention; at present, after-sale quality assurance of power batteries by most electric automobile manufacturers is defined as that the capacity of a power battery pack is attenuated to 80 percent, so that the power battery pack is regarded as the end of life and can be replaced;
therefore, under the condition that the types of the pure electric vehicles are consistent and the specifications of the power batteries are consistent, the problem that the pure electric vehicles need to be replaced under the condition of using the pure electric vehicles to any degree can be solved, and the reverse neural network construction can be carried out according to the corresponding technical parameters of the power batteries or the running parameters of the vehicles when the pure electric vehicles actually run to a critical state (the capacity is attenuated to 80%), so that the accurate evaluation index of the power batteries after being used for a period of time is obtained, and the evaluation index is suitable for the reference of most vehicles in the vehicles of the type, so that the actual application value is realized, and the rights of consumers are reasonably maintained on the premise of having good accuracy.
The existing methods for evaluating the performance or the predicted service life of the power battery are mostly neural network models, the models need to extract continuous parameters for modeling, and certain sequencing parameters cannot be used, for example, the modeling cannot be performed according to road conditions, and the influence of environmental factors such as environmental temperature, solar radiation intensity and the like on the battery and the service life of the vehicle cannot be effectively considered; in the current process of constructing the neural network model by multiple parameters, the parameters are manually considered when part of the parameters are related to the service life of the power battery, and corresponding scientific analysis is not performed, so that effective related analysis is lacking in the process of constructing the multiple parameters.
Disclosure of Invention
The invention provides a method for constructing an evaluation index of a power battery of a pure electric vehicle based on big data analysis.
The scheme of the invention is as follows:
the method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis comprises the following steps:
A1. the plurality of vehicles with the same type and the same matched power battery are respectively marked as P 1 、P 2 ……P n The vehicle is provided with a communication module for uploading data to a server through a communication device;
a2, P 1 In the running process of a vehicle, collecting the capacity attenuation rate C% of a vehicle power battery according to a certain sampling period T, and taking a plurality of C% collected in a plurality of sampling periods T as a calibration sample; synchronously collecting vehicle parameter index set V according to sampling period T ] 1 And power battery parameter index set [ B ]] 1 The method comprises the steps of carrying out a first treatment on the surface of the The [ V ]] 1 Comprising a plurality of vehicle-related parameters when the vehicle is traveling, said [ B ]] 1 Including a plurality of parameters related to a power battery mounted on the vehicle; the calibration sample, [ V ]] 1 、[B] 1 Uploading to a server;
a3, P 2 ……P n Repeating the step A2 to obtain a plurality of corresponding vehicle parameter index sets [ V ] when the values of C% are different] 2 ……[V] n And a plurality of power cell parameter index sets [ B ]] 2 ……[B] n ；
A4, pair [ V ]] 1 ……[V] n The plurality of parameters are statistically classified into a sequencing type parameter DX or a continuous type parameter DB, and [ B ]] 1 ……[B] n Statistically classifying a plurality of parameters of the plurality of parametersThe numbers are respectively divided into sequencing type parameters DX or continuous type parameters DB;
extracting n parameters of n vehicles in the same sampling period T for the parameter DB, checking the normal distribution of the n parameters, and eliminating the sampling point and n parameter values under the sampling point if the normal distribution is not met; when the distribution accords with normal distribution, the distribution is reserved; when the number value of the reserved acquisition points is more than 90% of the total acquisition points, defining the residual data in the corresponding parameter DB as DB ND Parameters are reserved;
a5, eliminating 1 st DB ND Parameters are according to 1% of data before and 99% of data after normal distribution, and 1% -99% of envelope data in normal distribution is reserved; repeating the step A5 by analogy, and repeating the step A5 for all the DB ND Processing parameters;
a6, taking the calibration sample as a target value and taking the 1 st DB ND Parameter data is used as input parameters, and a CNN convolutional neural network is used for establishing a neural network model MM 1 The method comprises the steps of carrying out a first treatment on the surface of the Analogize to a plurality of DB ND Obtaining parameters for multiple neural network models MM 2 ……MM n ；
A7, using neural network model MM 1 ……MM n The calculated value at each sampling point is taken as a dependent variable, and a calibration sample is taken as the independent variable; respectively carrying out correlation analysis between the calibration sample and calculated values of a plurality of neural network models by using a correlation analysis method; obtaining a correlation coefficient gamma and a saliency check probability rho%, extracting one or more neural network models with the saliency check probability rho% < 0.05 and the maximum correlation coefficient gamma value, and forming a continuous model group (MM)]；
A8, establishing a regression equation of the calculated value of a certain neural network model in the calibration sample and the continuous model set [ MM ] by using a regression analysis method; the number of the established regression equations is consistent with the number of the calculated values of the neural network model in the [ MM ];
a9, selecting m vehicles, predicting the numerical value of each C% in sequence according to a sampling period by using a regression equation in the step A8, comparing the prediction result with the C% in the actual running process of the m vehicles, and judging that the comprehensive error rate of the comparison result is less than or equal to 5% to be qualified; the integrated error rate is the root mean square value of the error rate at all sampling points;
And extracting one or more regression equations with the lowest comprehensive error rate from the regression equations as a continuous parameter prediction model.
As a preferable technical solution, the step A7 further includes the following steps:
a7-1, performing correlation analysis on the data in the 1 st sequencing-type parameter DX and the calculated values of all mathematical models MM in the continuous model group [ MM ] by using a correlation analysis method;
a7-2, if the correlation grade between the sequencing type parameter DX and more than 50% of mathematical models MM in the continuous model group [ MM ] is 3 grade or more, entering the next step;
a7-3, calculating the ratio of each sequencing data in the DX parameter corresponding to the record vehicle when the capacity attenuation rate C% of the power battery of the vehicle changes from 100% to 80% in the sampling period T; the ratio Z of the first sequenced data in the 1 st DX parameter 1-1 Ratio Z of% to second sequenced data 1-2 % … … Z-ordered data occupancy Z 1-z The%; calculating the occupation ratio of each sequencing data in the respective first DX parameters of the n vehicles respectively; forming a qxnxz matrix; wherein Q is the number of samplings;
a7-4, training by using the QXnXZ matrix as an input layer and the variation of the battery capacity attenuation rate C% from 100% to 80% as a target value and using a CNN convolutional neural network model to obtain a multilayer neural network model;
A7-5, repeating the step A7-1 with the residual sequencing parameters DX respectively; in the step A9, m vehicles are selected, the numerical value of each C% is predicted by the multi-layer neural network model according to the sampling frequency of intervals, the predicted result is compared with the C% in the actual running process of the m vehicles, and the comprehensive error rate of the compared result is less than or equal to 5% and is judged to be qualified;
and extracting one or more multi-layer neural network models with the lowest comprehensive error rate from the plurality of multi-layer neural network models as sequencing type parameter prediction models.
As a preferable embodiment, the vehicle parameter index set [ V]Comprising a vehicle accumulated driving mileage L and a vehicle accumulated using Time 1 Acceleration frequency, braking frequency and driving road condition; the acceleration frequency is acceleration frequency/within a single sampling period; the braking frequency is braking frequency/within a single sampling period;
the related index set [ B ] of the power battery]Includes cyclic charge times N1, accumulated charge Time Time 3 The method comprises the steps of carrying out a first treatment on the surface of the Open circuit voltage U1, operating voltage U2, internal resistance R1, self-discharge rate DeltaC; maximum accumulated working Time Time of single battery module 2 The method comprises the steps of carrying out a first treatment on the surface of the The open-circuit voltage U1 and the working voltage U2 are minimum values measured by a single battery module in a normal working state in the power battery pack; and the internal resistance R1 and the self-discharge rate delta C are the maximum measured value of the single battery module in the normal working state in the power battery pack.
As a preferable technical scheme, the vehicle parameter index set [ V]Accumulated mileage L in the vehicle and accumulated use Time of the vehicle 1 Acceleration frequency, braking frequency and related index set [ B ] of power battery]The number of cyclic charging times N1 and the accumulated charging Time Time 3 Maximum accumulated working Time Time of single battery module 2 Setting an open circuit voltage U1, an operating voltage U2, an internal resistance R1 and a self-discharge rate delta C as continuous parameters DB in the step A4; the vehicle parameter index set [ V ]]Setting the driving road condition in step A4 as a sequencing parameter DX; meanwhile, the environmental temperature parameter Temp-1 in the running state of the vehicle, the environmental temperature parameter Temp-2 in the parking state of the vehicle and the solar irradiation radiation intensity SH are also collected, and the Temp-1, temp-2 and SH are set as sequencing parameters DX in the step A4; the vehicle parameter index set [ V]The medium acceleration frequency is acceleration frequency/within a single sampling period, and the braking frequency is braking frequency/within a single sampling period.
As an optimal technical scheme, the driving road condition in the vehicle parameter index set [ V ] is divided into 4 sequencing data, which are urban congestion working conditions, urban general working conditions, suburban working conditions and high-speed working conditions, and the urban congestion working conditions are the vehicle parameter index set [ V ] when the braking frequency is more than 7.73; the brake frequency is less than or equal to 2.8 and less than 7.73, and the common urban working condition is adopted; the braking frequency is more than or equal to 1.82 and less than 2.8, which is suburban working condition; the braking frequency is less than 1.82 and is a high-speed working condition; the braking frequency is average braking frequency per kilometer;
The solar illumination radiation intensity SH is divided into 5 sequencing data, which are a first-class region, a second-class region, a third-class region and a fourth-class region;
the area is 3200-3300 hours of irradiation time per year, and the annual radiation quantity is 1860-2330 kW.h/m 2 ；
The second type of region is 3000-3200 hours of sunshine hours all year round, and the annual radiation is 1630-1860 kW.h/m 2 ；
The three areas are 2200-3000 hours of sunshine hours all year round, and the annual radiation quantity is 1390-1630 kW.h/m 2 ；
The four areas are 1400-2200 hours of sunshine hours all year round, and the annual radiation quantity is 1160-1390 kW.h/m 2 ；
Five types of regions: the number of sunshine hours in the whole year is about 1000-1400 hours, and the annual radiation quantity is 933-1160 kW.h/m 2 。
As a preferable technical scheme, the relevance grade in the step A7-1 is classified into the following 4 grades,
taking the data in the 1 st sequencing-type parameter DX as first column data and taking the calculated values of all mathematical models MM in the continuous model group [ MM ] as second column data;
the level 1 is that no correlation exists between the first column data and the second column data at all;
the level 2 is that fuzzy relativity exists between the first column data and the second column data;
the level 3 is that strong correlation exists between the first column data and the second column data, but no clear functional relation exists;
The level 4 is that a clear functional relation exists between the first column data and the second column data.
As a preferred embodiment, the correlation analysis method is a Spearman analysis method or a covariance analysis method.
As an preferable technical solution, the sampling period T in the step A2 is set according to one of a certain Time or a certain mileage Trip;
the sampling period T is set according to a certain mileage Trip, wherein the sampling period T is one of a driving mileage of 0.5 ten thousand kilometers, 1 ten thousand kilometers, 2 ten thousand kilometers and 2.5 ten thousand kilometers;
the sampling period T is set according to a certain Time Time, wherein the sampling period T is one of 3 months, 6 months, 8 months and 1 year.
As a preferable technical solution, the communication device in the step A1 is an internet communication device or a mobile communication device.
Due to the adoption of the technical scheme, the method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis comprises the following steps: A1. the plurality of vehicles with the same type and the same matched power battery are respectively marked as P 1 、P 2 ……P n The vehicle is provided with a communication module for uploading data to a server through a communication device;
a2, P 1 In the running process of a vehicle, collecting the capacity attenuation rate C% of a vehicle power battery according to a certain sampling period T, and taking a plurality of C% collected in a plurality of sampling periods T as a calibration sample; synchronously collecting vehicle parameter index set V according to sampling period T ] 1 And power battery parameter index set [ B ]] 1 The method comprises the steps of carrying out a first treatment on the surface of the The [ V ]] 1 Comprising a plurality of vehicle-related parameters when the vehicle is traveling, said [ B ]] 1 Including a plurality of parameters related to a power battery mounted on the vehicle; the calibration sample, [ V ]] 1 、[B] 1 Uploading to a server;
a3, P 2 ……P n Repeating the step A2 to obtain a plurality of corresponding vehicle parameter index sets [ V ] when the values of C% are different] 2 ……[V] n And a plurality of power cell parameter index sets [ B ]] 2 ……[B] n ；
A4, pair [ V ]] 1 ……[V] n The plurality of parameters are statistically classified into a sequencing type parameter DX or a continuous type parameter DB, and [ B ]] 1 ……[B] n Statistics of multiple parameters in (a)A learning classification for dividing the plurality of parameters into sequencing-type parameters DX or continuous-type parameters DB, respectively;
extracting n parameters of n vehicles in the same sampling period T for the parameter DB, checking the normal distribution of the n parameters, and eliminating the sampling point and n parameter values under the sampling point if the normal distribution is not met; when the distribution accords with normal distribution, the distribution is reserved; when the number value of the reserved acquisition points is more than 90% of the total acquisition points, defining the residual data in the corresponding parameter DB as DB ND Parameters are reserved;
a5, eliminating 1 st DB ND Parameters are according to 1% of data before and 99% of data after normal distribution, and 1% -99% of envelope data in normal distribution is reserved; repeating the step A5 by analogy, and repeating the step A5 for all the DB ND Processing parameters;
a6, taking the calibration sample as a target value and taking the 1 st DB ND Parameter data is used as input parameters, and a CNN convolutional neural network is used for establishing a neural network model MM 1 The method comprises the steps of carrying out a first treatment on the surface of the Analogize to a plurality of DB ND Obtaining parameters for multiple neural network models MM 2 ……MM n ；
A7, using neural network model MM 1 ……MM n The calculated value at each sampling point is taken as a dependent variable, and a calibration sample is taken as the independent variable; respectively carrying out correlation analysis between the calibration sample and calculated values of a plurality of neural network models by using a correlation analysis method; obtaining a correlation coefficient gamma and a saliency check probability rho%, extracting one or more neural network models with the saliency check probability rho% < 0.05 and the maximum correlation coefficient gamma value, and forming a continuous model group (MM)]；
A8, establishing a regression equation of the calculated value of a certain neural network model in the calibration sample and the continuous model set [ MM ] by using a regression analysis method; the number of the established regression equations is consistent with the number of the calculated values of the neural network model in the [ MM ];
a9, selecting m vehicles, predicting the numerical value of each C% in sequence according to a sampling period by using a regression equation in the step A8, comparing the prediction result with the C% in the actual running process of the m vehicles, and judging that the comprehensive error rate of the comparison result is less than or equal to 5% to be qualified; the integrated error rate is the root mean square value of the error rate at all sampling points;
And extracting one or more regression equations with the lowest comprehensive error rate from the regression equations as a continuous parameter prediction model.
The invention has the advantages that: firstly, carrying out normal distribution statistics on data collected by each vehicle under continuous parameters to enable the data to have statistical significance, then constructing a CNN convolutional neural network model on the collected data of a plurality of vehicles, and then reversely outputting a row of calculated values by using the neural network model to convert the collected data of the plurality of vehicles into a row of calculated values with statistical significance, wherein the calculated values have better prediction accuracy; and then, a regression analysis method in statistics is used for analyzing the correlation between the capacity attenuation rate C% of the vehicle power battery and the calculated value of each neural network model, a regression equation with obvious correlation is extracted to serve as a prediction model of the power battery performance or service life, the calculated amount is greatly reduced compared with the traditional neural network model calculation method, and compared with a single correlation regression prediction method, the accuracy and the adaptability are better, and the practical application value is wide.
Drawings
FIG. 1 is a schematic diagram of Tesla Model S/Model X20 ten thousand miles (about 32 ten thousand kilometers) battery capacity retention;
Fig. 2 is a schematic diagram of a method for constructing an evaluation index of a power battery of a pure electric vehicle in the first embodiment;
fig. 3 is a schematic diagram of a method for constructing an evaluation index of a power battery of a pure electric vehicle in a second embodiment.
Detailed Description
In order to make up for the defects, the invention provides a method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis, which solves the problems in the prior art.
A1. The plurality of vehicles with the same type and the same matched power battery are respectively marked as P 1 、P 2 ……P n The vehicle is provided with a communication module for uploading data to a server through a communication device;
a2, P 1 In the running process of a vehicle, collecting the capacity attenuation rate C% of a vehicle power battery according to a certain sampling period T, and taking a plurality of C% collected in a plurality of sampling periods T as a calibration sample; synchronously collecting vehicle parameter index set V according to sampling period T] 1 And power battery parameter index set [ B ]] 1 The method comprises the steps of carrying out a first treatment on the surface of the The [ V ]] 1 Comprising a plurality of vehicle-related parameters when the vehicle is traveling, said [ B ]] 1 Including a plurality of parameters related to a power battery mounted on the vehicle; the calibration sample, [ V ]] 1 、[B] 1 Uploading to a server;
a3, P 2 ……P n Repeating the step A2 to obtain a plurality of corresponding vehicle parameter index sets [ V ] when the values of C% are different ] 2 ……[V] n And a plurality of power cell parameter index sets [ B ]] 2 ……[B] n ；
A4, pair [ V ]] 1 ……[V] n The plurality of parameters are statistically classified into a sequencing type parameter DX or a continuous type parameter DB, and [ B ]] 1 ……[B] n The method comprises the steps of performing statistical classification on a plurality of parameters, and dividing the plurality of parameters into sequencing type parameters DX or continuous type parameters DB respectively;
extracting n parameters of n vehicles in the same sampling period T for the parameter DB, checking the normal distribution of the n parameters, and eliminating the sampling point and n parameter values under the sampling point if the normal distribution is not met; when the distribution accords with normal distribution, the distribution is reserved; when the number value of the reserved acquisition points is more than 90% of the total acquisition points, defining the residual data in the corresponding parameter DB as DB ND Parameters are reserved;
a5, eliminating 1 st DB ND Parameters are according to 1% of data before and 99% of data after normal distribution, and 1% -99% of envelope data in normal distribution is reserved; repeating the step A5 by analogy, and repeating the step A5 for all the DB ND Processing parameters;
a6, taking the calibration sample as a target value and taking the 1 st DB ND Parameter data is used as input parameters, and CNN convolutional neural network is used for building a neural networkModel MM 1 The method comprises the steps of carrying out a first treatment on the surface of the Analogize to a plurality of DB ND Obtaining parameters for multiple neural network models MM 2 ……MM n ；
A7, using neural network model MM 1 ……MM n The calculated value at each sampling point is taken as a dependent variable, and a calibration sample is taken as the independent variable; respectively carrying out correlation analysis between the calibration sample and calculated values of a plurality of neural network models by using a correlation analysis method; obtaining a correlation coefficient gamma and a saliency check probability rho%, extracting one or more neural network models with the saliency check probability rho% < 0.05 and the maximum correlation coefficient gamma value, and forming a continuous model group (MM)]；
A8, establishing a regression equation of the calculated value of a certain neural network model in the calibration sample and the continuous model set [ MM ] by using a regression analysis method; the number of the established regression equations is consistent with the number of the calculated values of the neural network model in the [ MM ];
a9, selecting m vehicles, predicting the numerical value of each C% in sequence according to a sampling period by using a regression equation in the step A8, comparing the prediction result with the C% in the actual running process of the m vehicles, and judging that the comprehensive error rate of the comparison result is less than or equal to 5% to be qualified; the integrated error rate is the root mean square value of the error rate at all sampling points;
and extracting one or more regression equations with the lowest comprehensive error rate from the regression equations as a continuous parameter prediction model.
As a preferable technical solution, the step A7 further includes the following steps:
a7-1, performing correlation analysis on the data in the 1 st sequencing-type parameter DX and the calculated values of all mathematical models MM in the continuous model group [ MM ] by using a correlation analysis method;
a7-2, if the correlation grade between the sequencing type parameter DX and more than 50% of mathematical models MM in the continuous model group [ MM ] is 3 grade or more, entering the next step;
a7-3, calculating the occupation of each sequencing data in the DX parameters corresponding to the record vehicle when the capacity attenuation rate C% of the power battery of the vehicle is changed from 100% to 80% in the sampling period TRatio of; the ratio Z of the first sequenced data in the 1 st DX parameter 1-1 Ratio Z of% to second sequenced data 1-2 % … … Z-ordered data occupancy Z 1-z The%; calculating the occupation ratio of each sequencing data in the respective first DX parameters of the n vehicles respectively; forming a qxnxz matrix; wherein Q is the number of samplings;
a7-4, training by using the QXnXZ matrix as an input layer and the variation of the battery capacity attenuation rate C% from 100% to 80% as a target value and using a CNN convolutional neural network model to obtain a multilayer neural network model;
a7-5, repeating the step A7-1 with the residual sequencing parameters DX respectively; in the step A9, m vehicles are selected, the numerical value of each C% is predicted by the multi-layer neural network model according to the sampling frequency of intervals, the predicted result is compared with the C% in the actual running process of the m vehicles, and the comprehensive error rate of the compared result is less than or equal to 5% and is judged to be qualified;
And extracting one or more multi-layer neural network models with the lowest comprehensive error rate from the plurality of multi-layer neural network models as sequencing type parameter prediction models.
As a preferable embodiment, the vehicle parameter index set [ V]Comprising a vehicle accumulated driving mileage L and a vehicle accumulated using Time 1 Acceleration frequency, braking frequency and driving road condition; the acceleration frequency is acceleration frequency/within a single sampling period; the braking frequency is braking frequency/within a single sampling period;
the related index set [ B ] of the power battery]Includes cyclic charge times N1, accumulated charge Time Time 3 The method comprises the steps of carrying out a first treatment on the surface of the Open circuit voltage U1, operating voltage U2, internal resistance R1, self-discharge rate DeltaC; maximum accumulated working Time Time of single battery module 2 The method comprises the steps of carrying out a first treatment on the surface of the The open-circuit voltage U1 and the working voltage U2 are minimum values measured by a single battery module in a normal working state in the power battery pack; and the internal resistance R1 and the self-discharge rate delta C are the maximum measured value of the single battery module in the normal working state in the power battery pack.
As a preferable technical scheme, the vehicle parameter index set [ V]The accumulated driving mileage L,Accumulated Time of vehicle use 1 Acceleration frequency, braking frequency and related index set [ B ] of power battery ]The number of cyclic charging times N1 and the accumulated charging Time Time 3 Maximum accumulated working Time Time of single battery module 2 Setting an open circuit voltage U1, an operating voltage U2, an internal resistance R1 and a self-discharge rate delta C as continuous parameters DB in the step A4; the vehicle parameter index set [ V ]]Setting the driving road condition in step A4 as a sequencing parameter DX; meanwhile, the environmental temperature parameter Temp-1 in the running state of the vehicle, the environmental temperature parameter Temp-2 in the parking state of the vehicle and the solar irradiation radiation intensity SH are also collected, and the Temp-1, temp-2 and SH are set as sequencing parameters DX in the step A4; the vehicle parameter index set [ V]The medium acceleration frequency is acceleration frequency/within a single sampling period, and the braking frequency is braking frequency/within a single sampling period.
As an optimal technical scheme, the driving road condition in the vehicle parameter index set [ V ] is divided into 4 sequencing data, which are urban congestion working conditions, urban general working conditions, suburban working conditions and high-speed working conditions, and the urban congestion working conditions are the vehicle parameter index set [ V ] when the braking frequency is more than 7.73; the brake frequency is less than or equal to 2.8 and less than 7.73, and the common urban working condition is adopted; the braking frequency is more than or equal to 1.82 and less than 2.8, which is suburban working condition; the braking frequency is less than 1.82 and is a high-speed working condition; the braking frequency is average braking frequency per kilometer;
The solar illumination radiation intensity SH is divided into 5 sequencing data, which are a first-class region, a second-class region, a third-class region and a fourth-class region;
the area is 3200-3300 hours of irradiation time per year, and the annual radiation quantity is 1860-2330 kW.h/m 2 ；
The second type of region is 3000-3200 hours of sunshine hours all year round, and the annual radiation is 1630-1860 kW.h/m 2 ；
The three areas are 2200-3000 hours of sunshine hours all year round, and the annual radiation quantity is 1390-1630 kW.h/m 2 ；
The four areas are 1400-2200 hours of sunshine hours all year round, and the annual radiation quantity is 1160-1390 kW.h/m 2 ；
Five types of regions: sunshine duration of whole yearThe annual radiation quantity is 933-1160 kW.h/m for 1000-1400 hours 2 。
As a preferable technical scheme, the relevance grade in the step A7-1 is classified into the following 4 grades,
taking the data in the 1 st sequencing-type parameter DX as first column data and taking the calculated values of all mathematical models MM in the continuous model group [ MM ] as second column data;
the level 1 is that no correlation exists between the first column data and the second column data at all;
the level 2 is that fuzzy relativity exists between the first column data and the second column data;
the level 3 is that strong correlation exists between the first column data and the second column data, but no clear functional relation exists;
The level 4 is that a clear functional relation exists between the first column data and the second column data.
The correlation analysis method is a Spearman analysis method or a covariance analysis method.
The sampling period T in the step A2 is set according to one of a certain Time Time or a certain mileage Trip;
the sampling period T is set according to a certain mileage Trip, wherein the sampling period T is one of a driving mileage of 0.5 ten thousand kilometers, 1 ten thousand kilometers, 2 ten thousand kilometers and 2.5 ten thousand kilometers;
the sampling period T is set according to a certain Time Time, wherein the sampling period T is one of 3 months, 6 months, 8 months and 1 year.
The communication device in the step A1 is an internet communication device or a mobile communication device.
The invention is further described in connection with the following embodiments in order to make the technical means, the creation features, the achievement of the purpose and the effect of the invention easy to understand.
Example 1
As shown in fig. 1, the method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis is characterized in that: the method comprises the following steps in sequence:
a1, respectively marking a plurality of vehicles with the same vehicle type and the same matched power battery as P 1 、P 2 ……P n The vehicle is provided with a communication module for uploading data to a server through the Internet or Bluetooth or mobile communication equipment;
a2, P 1 In the running process of a vehicle, collecting the capacity attenuation rate C% of a vehicle power battery according to a certain sampling period T, and taking a plurality of C% collected in a plurality of sampling periods T as a calibration sample; synchronously collecting vehicle parameter index set V according to sampling period T] 1 And Power Battery parameter index set [ B ]] 1 The method comprises the steps of carrying out a first treatment on the surface of the The [ V ]] 1 Comprising a plurality of vehicle-related parameters during running of the vehicle, said [ B ]] 1 Including a plurality of parameters related to a power battery mounted on the vehicle; the sampling period T is set according to a certain Time Time or a certain mileage Trip; calibration samples, [ V ]] 1 、[B] 1 Uploading to a server;
because the variation of the C percent of the power batteries of different manufacturers and different types is large, the power batteries of partial vehicles can be collected in advance and classified after the actual service life of the power batteries of the partial vehicles is prolonged,
for a part of power batteries with shorter service life, for example, the actual service life is less than 20 ten thousand kilometers or less than 3 years, the C% change is quicker, namely, the service life of the battery is shorter, the sampling period T can be 2k-5k kilometers, or 15-30 days are taken according to the time interval;
For a longer life power cell, for example, the actual life is greater than 40 kilometers or greater than 7 years; the sampling period can be 0.5 ten thousand or 1 ten thousand or 2 ten thousand or 2.5 ten thousand kilometers of driving mileage; the sampling period T may also be 3 months or 6 months or 8 months or 1 year.
The vehicle parameter index set [ V ]]At least comprises a vehicle accumulated driving mileage L and a vehicle accumulated using Time 1 Acceleration frequency (in a sub/single sampling period), braking frequency (in a sub/single sampling period), and driving road condition; for example, the acceleration frequency is acquired every 5k km, and for a pure electric vehicle, the frequent acceleration can lead the power battery to be frequently in a short-time high-current discharge state,the related data show that the service life of the power battery is inversely related to the working condition, namely the higher the acceleration frequency is, the shorter the service life of the battery is; likewise, the higher the braking frequency, the shorter the battery life.
In addition, the related data show that the congestion condition of the vehicle during running can also influence the service life of the power battery; therefore, the method takes the driving road condition as sequencing data to perform related collection;
the related index set [ B ] of the power battery]Includes cyclic charge times N1, accumulated charge Time Time 3 The method comprises the steps of carrying out a first treatment on the surface of the Open circuit voltage U1, operating voltage U2, internal resistance R1, self-discharge rate DeltaC; maximum accumulated working Time Time of single battery module 2 The method comprises the steps of carrying out a first treatment on the surface of the The open-circuit voltage U1 and the working voltage U2 are minimum values measured by a single battery module in a normal working state in the power battery pack; and the internal resistance R1 and the self-discharge rate delta C are the maximum measured value of the single battery module in the normal working state in the power battery pack.
A3, P 2 ……P n Repeating the step A2 to obtain a plurality of corresponding vehicle parameter index sets [ V ] when the values of C% are different] 2 ……[V] n And a plurality of power cell parameter index sets [ B ]] 2 ……[B] n ；
A4, pair [ V ]] 1 ……[V] n The plurality of parameters are statistically classified into a sequencing type parameter DX or a continuous type parameter DB, and [ B ]] 1 ……[B] n The method comprises the steps of performing statistical classification on a plurality of parameters, and dividing the plurality of parameters into sequencing type parameters DX or continuous type parameters DB respectively;
the accumulated driving mileage L and the accumulated using Time of the vehicle 1 Acceleration frequency (in Time/single sampling period), braking frequency (in Time/single sampling period), cyclic charge number N1, accumulated charge Time Time 3 Maximum accumulated working Time Time of single battery module 2 The method comprises the steps of carrying out a first treatment on the surface of the The open circuit voltage U1, the working voltage U2, the internal resistance R1 and the self-discharge rate delta C are set as continuous parameters DB in the step A4; the continuous parameter DX is fixed ratio data in statistics, such as open circuit voltage 48V of single battery module as reference, and is measured subsequently The amount is 48.1V, 48.0V and 39.9V;
the driving road condition, the environmental temperature parameter Temp-1 in the driving state of the vehicle, the environmental temperature parameter Temp-2 in the parking state of the vehicle and the solar radiation intensity SH are set as sequencing parameters DX in the step A4;
the sequencing parameter DX is sequencing data in statistics, for example, the environmental temperature parameter Temp-1 in a vehicle running state can be divided into 5 or 7 grades according to the temperature range according to the thermal comfort induction of a human body, and the environmental temperature parameter Temp-2 in a vehicle parking state can be divided into 3 or 5 grades according to the self-discharge rate deltac of the power battery at different temperatures;
in a recommended division manner, the driving road condition is divided into 4 sequencing data: urban congestion working conditions, urban general working conditions, suburban working conditions and high-speed working conditions, wherein the urban congestion working conditions (average braking times per kilometer) are adopted when the braking frequency in the vehicle parameter index set [ V ] is more than 7.73; the brake frequency is less than or equal to 2.8 and less than 7.73, and the common urban working condition is adopted; the braking frequency is more than or equal to 1.82 and less than 2.8, which is suburban working condition; the braking frequency is less than 1.82 and is a high-speed working condition;
in a recommended partitioning, the solar radiation intensity SH is divided into 5 sequencing data:
A class of regions: the total number of irradiation times per year is 3200-3300 hours, and the annual radiation quantity is 1860-2330 kW.h/m 2 ；
Two types of regions: the sunshine duration of the whole year is 3000-3200 hours, and the annual radiation quantity is 1630-1860 kW.h/m 2 ；
Three types of regions: the number of sunshine hours in the whole year is 2200-3000 hours, and the annual radiation quantity is 1390-1630kW.h/m 2 ；
Four types of regions: the sunshine hours of the whole year is 1400-2200 hours, and the annual radiation quantity is 1160-1390 kW.h/m 2 ；
Five types of regions: the number of sunshine hours in the whole year is about 1000-1400 hours, and the annual radiation quantity is 933-1160 kW.h/m 2 ；
For the DB parameter, since the parameter is a continuous parameter, for a plurality of vehicles, the numerical value of the parameter basically accords with normal distribution, and if the data at a certain sampling point is abnormal, the data is rejected so as to ensure the accuracy of the subsequent model construction.
N parameters of n vehicles in the same sampling period T are extracted, normal distribution of the n parameters is checked, and statistical processing software such as SPSS and the like can be directly used for analysis; if the normal distribution is not met, eliminating the sampling point and n parameter values under the sampling point; if the normal distribution is met, reserving; when the number value of the reserved acquisition points is more than 90% of the total acquisition points, defining the residual data in the corresponding DB parameters as DB ND Parameters are reserved; DB at this time ND The data in the parameters conform to normal distribution;
a5, eliminating 1 st DB ND The parameters are kept according to 1% before and 99% after normal distribution, i.e. the DB is maintained ND Data with normal distribution range between 1% and 99% in parameters, namely 98% satisfaction; when the data amount is large, a satisfaction degree of 95% or 90% can be set; repeating the step A5 by analogy, and repeating the step A5 for all the DB ND Processing parameters;
a6, taking the calibration sample as a target value, wherein the calibration sample at the moment is a numerical value which keeps C% of all sampling points, and the 1 st DB ND Parameter data as input parameters, DB ND Partial sampling point parameter is missing in the parameter; building neural network model MM using CNN convolutional neural network 1 The method comprises the steps of carrying out a first treatment on the surface of the MM according to neural network model 1 Input DB ND When the parameters are used, the change value of the capacity attenuation rate C% of the vehicle power battery can be obtained, namely, the prediction function is realized, and the change value of C% can be reversely input to obtain the MM 1 Calculated value of model at each sampling point, the MM 1 The data of the calculated values of all the sampling points calculated in the reverse direction of the model are complete and have statistical significance, so the calculated values are used when the relevance is calculated later; the method comprises the steps that through the construction of a neural network model, data values of a plurality of vehicles are simplified into single-column data with statistical significance; when the subsequent correlation analysis is carried out with the calibration sample, two columns of data are adopted, and a plurality of correlation analysis methods can be conveniently used.
Analogize in the next place, howThe DB ND Obtaining parameters for multiple neural network models MM 2 ……MM n ；
A7, using neural network model MM 1 ……MM n The calculated value at each sampling point is taken as a dependent variable, and a calibration sample is taken as the independent variable; calibration samples and a plurality of neural network models MM are subjected to correlation analysis 1 ……MM n Respectively carrying out correlation analysis between the calculated values; obtaining a correlation coefficient gamma and a saliency check probability rho%, extracting one or more neural network models with the saliency check probability rho% < 0.05 and the maximum correlation coefficient gamma value to form a continuous model group (MM)]；
A8, establishing a regression equation of the calculated value of a certain neural network model in the calibration sample and the continuous model set [ MM ] by using a regression analysis method; the number of the established regression equations is consistent with the number of the calculated values of the neural network model in the [ MM ];
a9, selecting m vehicles, predicting the numerical value of each C% in sequence according to a sampling period by using a regression equation in the step A8, comparing the prediction result with the C% in the actual running process of the m vehicles, and judging that the comprehensive error rate of the comparison result is less than or equal to 5% to be qualified; the integrated error rate is the root mean square value of the error rate at all sampling points;
and extracting one or more regression equations with the lowest comprehensive error rate from the regression equations as a continuous parameter prediction model.
In the model construction process, firstly, normal distribution statistics is carried out on data collected by each vehicle under continuous parameters to enable the data to have statistical significance, then, a CNN convolutional neural network model is constructed on the collected data of a plurality of vehicles, then, a row of calculated values are reversely output by using the neural network model, the collected data of the plurality of vehicles are converted into a row of calculated values with statistical significance, and the calculated values have good prediction accuracy; and then, a regression analysis method in statistics is used for analyzing the correlation between the capacity attenuation rate C% of the vehicle power battery and the calculated value of each neural network model, a regression equation with obvious correlation is extracted to serve as a prediction model of the power battery performance or service life, the calculated amount is greatly reduced compared with the traditional neural network model calculation method, and compared with a single correlation regression prediction method, the accuracy and the adaptability are better, and the practical application value is wide.
Example two
As shown in fig. 2, in the conventional power battery evaluation index construction process, distance data or fixed ratio data, that is, variable data with continuity is used for calculation, but in practice, parameters of certain sequencing data types also have a certain influence on the performance of the power battery, if the continuous variable mode is used for calculation, the calculation amount is very large, so that the data is generally simplified into a plurality of sequences, that is, classification, such as environment temperature, external environment such as road condition, solar illumination intensity, and vehicle factors;
In the scheme, the continuous parameters DB are collected, the sequencing parameters DX are collected, and the two parameters are used for evaluating the power battery together, so that the actual use working condition of the power battery can be better reflected.
In the continuous parameter DB calculation process, correlation with a standard sample is carried out through correlation analysis; therefore, the sequencing parameters DX and the continuous model group [ MM ] which are subjected to correlation analysis, and then the power battery is evaluated together;
in the step A7, the method further includes the following steps:
a7-1, performing correlation analysis on the data in the 1 st sequencing-type parameter DX and the calculated values of all mathematical models MM in the continuous model group [ MM ] by using a correlation analysis method;
the correlation analysis may be performed using Spearman analysis, or covariance analysis, or other methods in SPSS software that may be used to correlate high-measurement sequencing data with fixed ratio data.
In the result of the correlation analysis, the correlation grade in the step A7-1 is classified into the following 4 grades,
taking the data in the 1 st sequencing-type parameter DX as first column data and taking the calculated values of all mathematical models MM in the continuous model group [ MM ] as second column data;
The level 1 is that no correlation exists between the first column data and the second column data at all;
the level 2 is that fuzzy relativity exists between the first column data and the second column data;
the level 3 is that strong correlation exists between the first column data and the second column data, but no clear functional relation exists;
the level 4 is that a clear functional relation exists between the first column data and the second column data.
A7-2, if the correlation level between the sequencing-type parameter DX and more than 50% of the mathematical models MM in the continuous model group [ MM ] is 3-level (strong correlation) or more; then it is indicated that there is a significant correlation of this sequenced parameter DX to the power cell capacity decay rate C%;
further calculations are made for this sequenced type parameter DX:
a7-3, calculating the ratio of each sequencing data in the DX parameter corresponding to the record vehicle when the capacity attenuation rate C% of the power battery of the vehicle changes from 100% to 80% in the sampling period T; the ratio Z of the first sequenced data in the 1 st DX parameter 1-1 Ratio Z of% to second sequenced data 1-2 % … … Z-ordered data occupancy Z 1-z The%; for example, a vehicle running condition is recorded once for 5000km of the vehicle, and the ratio of each running condition recorded with 3 sampling points is shown in table 1; the Z-th meaning belongs to a mathematical common expression method.
Table 1 table for collecting running condition of some vehicle every 5000km
Calculating the occupation ratio of each sequencing data in the respective first DX parameters of the n vehicles respectively; forming a qxnxz matrix; wherein Q is the number of sampling points;
a7-4, training by using the QXnXZ matrix as an input layer and the variation of the battery capacity attenuation rate C% from 100% to 80% as a target value and using a CNN convolutional neural network model to obtain a multilayer neural network model;
a7-5, repeating the step A7-1 with the residual sequencing parameters DX respectively; in the step A9, m vehicles are selected, the numerical value of each C% is predicted by the multi-layer neural network model according to the sampling period, the predicted result is compared with the C% in the actual running process of the m vehicles, and the comprehensive error rate of the compared result is less than or equal to 5% and is judged to be qualified;
and extracting one or more multi-layer neural network models with the lowest comprehensive error rate from the plurality of multi-layer neural network models as sequencing type parameter prediction models.
In the second embodiment, on the basis of outputting the continuous parameter prediction model, the sequencing parameter prediction model is also output at the same time, so that the application range of the power battery index evaluation method is greatly expanded, and the accuracy is further improved; and for a certain type of vehicle with huge maintenance and single use environment, the sequencing type parameter prediction model can be directly used for prediction, so that the prediction calculation process is greatly simplified.
The foregoing has shown and described the basic principles, main features and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (9)
1. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis is characterized by comprising the following steps:
A1. the plurality of vehicles with the same type and the same matched power battery are respectively marked as P 1 、P 2 ……P n The vehicle is provided with a communication module for uploading data to a server through a communication device;
a2, P 1 In the running and using process of the vehicle,collecting the capacity attenuation rate C% of the vehicle power battery according to a certain sampling period T, and taking a plurality of C% collected by a plurality of sampling periods T as a calibration sample; synchronously collecting vehicle parameter index set V according to sampling period T] 1 And power battery parameter index set [ B ] ] 1 The method comprises the steps of carrying out a first treatment on the surface of the The [ V ]] 1 Comprising a plurality of vehicle-related parameters when the vehicle is traveling, said [ B ]] 1 Including a plurality of parameters related to a power battery mounted on the vehicle; the calibration sample, [ V ]] 1 、[B] 1 Uploading to a server;
a3, P 2 ……P n Repeating the step A2 to obtain a plurality of corresponding vehicle parameter index sets [ V ] when the values of C% are different] 2 ……[V] n And a plurality of power cell parameter index sets [ B ]] 2 ……[B] n ；
A4, pair [ V ]] 1 ……[V] n The plurality of parameters are statistically classified into a sequencing type parameter DX or a continuous type parameter DB, and [ B ]] 1 ……[B] n The plurality of parameters in the sequence type parameter DX or the continuous type parameter DB are classified into the sequence type parameter DX or the continuous type parameter DB in a statistical manner;
extracting n parameters of n vehicles in the same sampling period T for the parameter DB, checking the normal distribution of the n parameters, and eliminating the sampling point and n parameter values under the sampling point if the normal distribution is not met; when the distribution accords with normal distribution, the distribution is reserved; when the number value of the reserved acquisition points is more than 90% of the total acquisition points, defining the residual data in the corresponding parameter DB as DB ND Parameters are reserved;
a5, eliminating 1 st DB ND Parameters are according to 1% of data before and 99% of data after normal distribution, and 1% -99% of envelope data in normal distribution is reserved; repeating the step A5 by analogy, and repeating the step A5 for all the DB ND Processing parameters;
a6, taking the calibration sample as a target value and taking the 1 st DB ND Parameter data is used as input parameters, and a CNN convolutional neural network is used for establishing a neural network model MM 1 The method comprises the steps of carrying out a first treatment on the surface of the Analogize to a plurality of DB ND Obtaining parameters for multiple neural network models MM 2 ……MM n ；
A7, using neural network model MM 1 ……MM n The calculated value at each sampling point is taken as a dependent variable, and a calibration sample is taken as the independent variable; respectively carrying out correlation analysis between the calibration sample and calculated values of a plurality of neural network models by using a correlation analysis method; obtaining a correlation coefficient gamma and a saliency check probability rho%, extracting one or more neural network models with the saliency check probability rho% < 0.05 and the maximum correlation coefficient gamma value, and forming a continuous model group (MM)]；
A8, establishing a regression equation of the calculated value of a certain neural network model in the calibration sample and the continuous model set [ MM ] by using a regression analysis method; the number of the established regression equations is consistent with the number of the calculated values of the neural network model in the [ MM ];
a9, selecting m vehicles, predicting the numerical value of each C% in sequence according to a sampling period by using a regression equation in the step A8, comparing the prediction result with the C% in the actual running process of the m vehicles, and judging that the comprehensive error rate of the comparison result is less than or equal to 5% to be qualified; the integrated error rate is the root mean square value of the error rate at all sampling points;
And extracting one or more regression equations with the lowest comprehensive error rate from the regression equations as a continuous parameter prediction model.
2. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 1, wherein the step A7 further comprises the following steps:
a7-1, performing correlation analysis on the data in the 1 st sequencing-type parameter DX and the calculated values of all mathematical models MM in the continuous model group [ MM ] by using a correlation analysis method;
a7-2, if the correlation grade between the sequencing type parameter DX and more than 50% of mathematical models MM in the continuous model group [ MM ] is 3 grade or more, entering the next step;
a7-3, calculating and recording each sequencing number in the DX parameter corresponding to the vehicle when the capacity attenuation rate C% of the power battery of the vehicle is changed from 100% to 80% in the sampling period TThe ratio of the data; the ratio Z of the first sequenced data in the 1 st DX parameter 1-1 Ratio Z of% to second sequenced data 1-2 % … … Z-ordered data occupancy Z 1-z The%; calculating the occupation ratio of each sequencing data in the respective first DX parameters of the n vehicles respectively; forming a qxnxz matrix; wherein Q is the number of samplings;
a7-4, training by using the QXnXZ matrix as an input layer and the variation of the battery capacity attenuation rate C% from 100% to 80% as a target value and using a CNN convolutional neural network model to obtain a multilayer neural network model;
A7-5, repeating the step A7-1 with the residual sequencing parameters DX respectively; in the step A9, m vehicles are selected, the numerical value of each C% is predicted by the multi-layer neural network model according to the sampling frequency of intervals, the predicted result is compared with the C% in the actual running process of the m vehicles, and the comprehensive error rate of the compared result is less than or equal to 5% and is judged to be qualified;
and extracting one or more multi-layer neural network models with the lowest comprehensive error rate from the plurality of multi-layer neural network models as sequencing type parameter prediction models.
3. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 1, wherein the method comprises the following steps: the vehicle parameter index set [ V]Comprising a vehicle accumulated driving mileage L and a vehicle accumulated using Time 1 Acceleration frequency, braking frequency and driving road condition; the acceleration frequency is acceleration frequency/within a single sampling period; the braking frequency is braking frequency/within a single sampling period;
the related index set [ B ] of the power battery]Includes cyclic charge times N1, accumulated charge Time Time 3 The method comprises the steps of carrying out a first treatment on the surface of the Open circuit voltage U1, operating voltage U2, internal resistance R1, self-discharge rate DeltaC; maximum accumulated working Time Time of single battery module 2 The method comprises the steps of carrying out a first treatment on the surface of the The open-circuit voltage U1 and the working voltage U2 are minimum values measured by a single battery module in a normal working state in the power battery pack; the internal resistance R1 and the self-discharge rate delta C are measured by a single battery module in a normal working state in the power battery packThe maximum value of the quantity.
4. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 3, wherein the method comprises the following steps: the vehicle parameter index set [ V ]]Accumulated mileage L in the vehicle and accumulated use Time of the vehicle 1 Acceleration frequency, braking frequency and related index set [ B ] of power battery]The number of cyclic charging times N1 and the accumulated charging Time Time 3 Maximum accumulated working Time Time of single battery module 2 Setting an open circuit voltage U1, an operating voltage U2, an internal resistance R1 and a self-discharge rate delta C as continuous parameters DB in the step A4; the vehicle parameter index set [ V ]]Setting the driving road condition in step A4 as a sequencing parameter DX; meanwhile, the environmental temperature parameter Temp-1 in the running state of the vehicle, the environmental temperature parameter Temp-2 in the parking state of the vehicle and the solar irradiation radiation intensity SH are also collected, and the Temp-1, temp-2 and SH are set as sequencing parameters DX in the step A4; the vehicle parameter index set [ V ]The medium acceleration frequency is acceleration frequency/within a single sampling period, and the braking frequency is braking frequency/within a single sampling period.
5. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 1 or 4, wherein the method comprises the following steps: the driving road condition in the vehicle parameter index set [ V ] is divided into 4 sequencing data which are urban congestion working conditions, urban general working conditions, suburban working conditions and high-speed working conditions, and the vehicle parameter index set [ V ] is the urban congestion working conditions when the braking frequency in the vehicle parameter index set [ V ] is more than 7.73; the brake frequency is less than or equal to 2.8 and less than 7.73, and the common urban working condition is adopted; the braking frequency is more than or equal to 1.82 and less than 2.8, which is suburban working condition; the braking frequency is less than 1.82 and is a high-speed working condition; the braking frequency is average braking frequency per kilometer;
the solar illumination radiation intensity SH is divided into 5 sequencing data, which are a first-class region, a second-class region, a third-class region and a fourth-class region;
the area is 3200-3300 hours of irradiation time per year, and the annual radiation quantity is 1860-2330 kW.h/m 2 ；
The second type region is the sunshine of whole yearThe number of the radiation is 3000-3200 hours, and the annual radiation is 1630-1860 kW.h/m 2 ；
The three areas are 2200-3000 hours of sunshine hours all year round, and the annual radiation quantity is 1390-1630 kW.h/m 2 ；
The four areas are 1400-2200 hours of sunshine hours all year round, and the annual radiation quantity is 1160-1390 kW.h/m 2 ；
Five types of regions: the number of sunshine hours in the whole year is about 1000-1400 hours, and the annual radiation quantity is 933-1160 kW.h/m 2 。
6. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 2, wherein the method comprises the following steps: the relevance grade in step A7-1 is classified into the following 4 grades,
taking the data in the 1 st sequencing-type parameter DX as first column data and taking the calculated values of all mathematical models MM in the continuous model group [ MM ] as second column data;
the level 1 is that no correlation exists between the first column data and the second column data at all;
the level 2 is that fuzzy relativity exists between the first column data and the second column data;
the level 3 is that strong correlation exists between the first column data and the second column data, but no clear functional relation exists;
the level 4 is that a clear functional relation exists between the first column data and the second column data.
7. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 2, wherein the method comprises the following steps: the correlation analysis method is a Spearman analysis method or a covariance analysis method.
8. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 1, wherein the method comprises the following steps: the sampling period T in the step A2 is set according to one of a certain Time Time or a certain mileage Trip;
the sampling period T is set according to a certain mileage Trip, wherein the sampling period T is one of a driving mileage of 0.5 ten thousand kilometers, 1 ten thousand kilometers, 2 ten thousand kilometers and 2.5 ten thousand kilometers;
the sampling period T is set according to a certain Time Time, wherein the sampling period T is one of 3 months, 6 months, 8 months and 1 year.
9. The method for constructing the evaluation index of the power battery of the pure electric vehicle based on big data analysis as claimed in claim 1, wherein the method comprises the following steps: the communication device in the step A1 is an internet communication device or a mobile communication device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160821.0A CN116381500A (en) | 2023-02-23 | 2023-02-23 | Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160821.0A CN116381500A (en) | 2023-02-23 | 2023-02-23 | Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116381500A true CN116381500A (en) | 2023-07-04 |
Family
ID=86962356
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310160821.0A Pending CN116381500A (en) | 2023-02-23 | 2023-02-23 | Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116381500A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117434463A (en) * | 2023-09-21 | 2024-01-23 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Method, device, equipment and storage medium for evaluating remaining life of power battery |
-
2023
- 2023-02-23 CN CN202310160821.0A patent/CN116381500A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117434463A (en) * | 2023-09-21 | 2024-01-23 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Method, device, equipment and storage medium for evaluating remaining life of power battery |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021185308A1 (en) | Online determination method and system for state of health of power battery pack of electric vehicle | |
| CN110570014B (en) | Electric vehicle charging load prediction method based on Monte Carlo and deep learning | |
| CN111301426B (en) | Method for predicting energy consumption in future driving process based on GRU network model | |
| CN112433169B (en) | Cloud power battery health degree evaluation system and method | |
| CN111381170A (en) | Electric vehicle battery pack health state prediction method and system based on big data | |
| CN110673039B (en) | Lithium iron phosphate battery SOC charging online correction method based on big data | |
| WO2022213597A1 (en) | New energy vehicle eic system safety feature database construction method | |
| CN109697854A (en) | The urban road traffic state assessment method of various dimensions | |
| CN115327422A (en) | Electric bus power battery health degree evaluation method based on charging and discharging behaviors | |
| CN109508826B (en) | Electric vehicle cluster schedulable capacity prediction method based on gradient lifting decision tree | |
| CN112986830A (en) | Lithium battery capacity estimation method based on convolution time memory neural network | |
| CN113765179B (en) | Cloud battery pack consistency analysis method based on hierarchical clustering and similar center distances | |
| CN113158345A (en) | New energy vehicle power battery capacity prediction method and system | |
| CN116381500A (en) | Method for constructing evaluation index of power battery of pure electric vehicle based on big data analysis | |
| CN114397592A (en) | Health degree prediction method and device for power battery of electric vehicle | |
| CN114976306A (en) | Intelligent echelon utilization method of power lithium battery based on vehicle-connected cloud | |
| CN115219903A (en) | Battery self-discharge rate abnormity judgment method and device based on Internet of vehicles data analysis | |
| CN115754372A (en) | Variable-temperature circulating lithium ion battery aging test method | |
| CN114646888A (en) | Assessment method and system for capacity attenuation of power battery | |
| CN117272783A (en) | Power battery life prediction method based on cloud edge collaborative multi-model fusion | |
| CN117665620A (en) | New energy automobile data-based battery health evaluation method | |
| CN115544900B (en) | Method for analyzing electric vehicle endurance mileage influence factors based on shape algorithm | |
| CN111562513A (en) | Power battery calendar life estimation method | |
| CN113128854B (en) | New energy automobile operation safety performance inspection method, device and system | |
| CN114545276A (en) | Power battery service life prediction method based on capacity test and Internet of vehicles big data |
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 |