CN112505570A - Method for estimating battery health state of electric automobile - Google Patents
Method for estimating battery health state of electric automobile Download PDFInfo
- Publication number
- CN112505570A CN112505570A CN202011411897.9A CN202011411897A CN112505570A CN 112505570 A CN112505570 A CN 112505570A CN 202011411897 A CN202011411897 A CN 202011411897A CN 112505570 A CN112505570 A CN 112505570A
- Authority
- CN
- China
- Prior art keywords
- battery
- state
- health
- data
- model
- 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
- 230000036541 health Effects 0.000 title claims abstract description 197
- 238000000034 method Methods 0.000 title claims abstract description 48
- 230000002159 abnormal effect Effects 0.000 claims abstract description 69
- 230000008929 regeneration Effects 0.000 claims abstract description 65
- 238000011069 regeneration method Methods 0.000 claims abstract description 65
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000012360 testing method Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 230000001419 dependent effect Effects 0.000 claims description 6
- 125000004122 cyclic group Chemical group 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012935 Averaging Methods 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 abstract description 5
- 238000013178 mathematical model Methods 0.000 abstract description 3
- 230000003862 health status Effects 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 208000032953 Device battery issue Diseases 0.000 description 1
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004836 empirical method Methods 0.000 description 1
- 229910001416 lithium ion Inorganic materials 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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]
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
- Tests Of Electric Status Of Batteries (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention provides a method for estimating the battery health state of an electric automobile, which comprises the following steps: the method comprises the steps of firstly collecting battery health state parameters of the same model by using a big data platform to obtain a database, establishing a regression model according to the battery health state parameter data in the database, secondly dividing output data of the regression model into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset threshold, substituting the capacity regeneration points and the processed abnormal point mean values into a first prediction model to generate a capacity regeneration result, substituting the normal points into a second prediction model to generate a normal result after data distribution balance processing, and finally generating a combination result which is output as an estimation result of the battery health state of the electric vehicle. The whole method flow can be realized by establishing a mathematical model, the influence of the capacity regeneration phenomenon is comprehensively considered, and the influence of chemical reaction on the battery capacity estimation is avoided, so that higher prediction precision is achieved.
Description
Technical Field
The invention relates to the technical field of batteries of electric vehicles, in particular to a method for estimating the health state of a battery of an electric vehicle.
Background
Battery state of health describes the state of health of a battery compared to a fresh battery. Battery state of health is typically defined by the ratio of the maximum capacity of the current cycle to the rated capacity. Generally, the battery failure threshold is recommended to be around 80% of rated capacity, and after the 80% threshold is exceeded, the battery capacity degradation shows an exponential decay trend. Therefore, when the state of health of the battery drops below 80%, the battery performance will deteriorate rapidly. In other words, a functional failure always occurs after the battery state of health is below the failure threshold. Lithium ion batteries of electric vehicles are considered unhealthy once the battery health is less than a predetermined fault threshold. However, the current battery state of health estimation of the electric vehicle still cannot accurately reflect the real capacity of the battery. To improve the true value of the battery capacity, the accuracy of the battery state of health estimation becomes more and more important.
The existing methods for evaluating the state of health of a battery are as follows: the enterprise big data platform collects operation data of various working conditions of different vehicle types, forms a cloud platform for data storage, and is used for monitoring the operation state of the electric vehicle by enterprises. A model for assessing battery state of health may be established by calling data from the big data platform. And the electrochemical model of the battery is based on the internal characteristics of the battery. When the working conditions (such as temperature and power) are changed, the required complex coupled nonlinear partial differential equation is difficult to solve. Although the equivalent circuit model may approximately represent the dynamic and static characteristics of the battery. However, the model is built up on impedance data, which is not experimentally available. The model can only approximate to the static characteristic and the dynamic characteristic of the battery, and the solving precision is not high.
In addition, the traditional model does not deeply consider the capacity regeneration phenomenon in the estimation of the battery health state, and the capacity regeneration phenomenon can influence the macroscopic driving mileage of the electric automobile, so that the mental pain or worry caused by worry of sudden power failure when a driver drives the electric automobile is caused, and the more accurate estimation of the battery health state can bring better driving experience.
Disclosure of Invention
The invention aims to solve the problem that the prediction accuracy of the estimation result of the battery health state is not high in the prior art.
In order to solve the above problem, an embodiment of the present invention discloses a method for estimating a state of health of a battery of an electric vehicle, including:
s1: acquiring battery health state parameters of batteries of the same type by using a big data platform, acquiring a database of the battery health state parameters, and establishing a regression model according to the battery health state parameter data in the database;
s2: analyzing the battery health state parameters by using a regression model to obtain output data of the regression model; the output data of the regression model comprises corrected battery health state information of the electric automobile;
s3: dividing output data into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset battery health state data threshold, and carrying out mean value processing on the abnormal points; substituting the capacity regeneration point and the processed abnormal point average value into a first prediction model to generate a capacity regeneration result;
s4: carrying out data distribution balance processing on the normal points, and substituting the processed normal points into a second prediction model to generate a normal result;
s5: and replacing the normal result at the same moment with the capacity regeneration result to generate a combined result, and outputting the combined result as an estimation result of the battery health state of the electric automobile.
By adopting the scheme, the battery health state parameters of the batteries with the same model in the big data platform are collected to establish a database of the battery health state parameters, then a regression model is established by utilizing the battery health parameter data in the database, and the data in the model is analyzed and trained to divide the output data of the regression model into outliers and normal points. And then classifying the outliers into capacity regeneration points and abnormal points, substituting the capacity regeneration points and the abnormal points subjected to mean value processing into a first prediction model, performing data distribution processing on the normal points, substituting the normal points into a second prediction model, and finally generating a combination result and outputting the combination result as an estimation result of the battery health state of the electric automobile. In the whole process of the method, no matter the battery health state parameters are collected, the regression model is built, or the output data of the regression model is analyzed and processed to generate a combined result, the method can be realized by building a mathematical model or software, actual measurement is not needed, and the operation is convenient. Moreover, the influence of the capacity regeneration phenomenon is comprehensively considered when the battery health state is estimated, and the influence of chemical reaction on the battery capacity estimation is avoided, so that the prediction accuracy of the battery health state is higher.
According to another embodiment of the present invention, in the method for estimating the battery state of health of an electric vehicle disclosed in the embodiment of the present invention, the battery state of health parameter includes a battery temperature, a battery current, a battery voltage, and initial battery state of health information.
According to another specific embodiment of the present invention, in the method for estimating the battery state of health of an electric vehicle disclosed in the embodiment of the present invention, in step S1, the establishing a regression model according to the battery state of health parameter data in the database includes: the method comprises the steps of dividing battery health state parameter data in a database into a training set and a testing set, establishing an initial model according to the training set, and verifying the initial model through the testing set to obtain a regression model.
By adopting the scheme, the battery temperature, the battery current, the battery voltage and the initial battery health state information are used as the input of the regression model, the model is trained and the regression model is obtained by dividing the battery health state parameter data into the training set and the testing set, and the corrected battery health state information is output through the regression model, so that the obtained output data is more accurate. In addition, the regression model obtained through the training can accurately represent the distribution trend of the corrected battery health state, and the subsequent operation of classifying the output data of the regression model is facilitated.
According to another embodiment of the present invention, in the method for estimating the state of health of a battery of an electric vehicle disclosed in the embodiment of the present invention, the regression model is a linear regression model; moreover, the first prediction model is an extreme gradient lifting model; the second prediction model is any one of a self-attention model, a long-short term memory network, a gated cyclic unit network, and a recurrent neural network.
According to another embodiment of the present invention, the method for estimating the battery state of health of an electric vehicle according to the embodiment of the present invention, in step S3, the dividing the output data into outliers and normal points includes: obtaining model parameters of a regression model; the model parameters comprise independent variables, dependent variables and constants; and obtaining a dividing line according to the independent variable, the dependent variable and the adjusted constant by adjusting the value of the constant, and dividing the output data of the regression model into outliers and normal points by using the dividing line.
According to another embodiment of the present invention, the method for estimating the battery health status of an electric vehicle, in step S3, the dividing the outlier into a capacity regeneration point and an abnormal point according to a preset threshold of the battery health status data includes:
s31: acquiring a battery health state value at the previous moment of each outlier and a battery health state value at the next moment of the outlier, and calculating a battery health state mean value of the outlier according to the battery health state value at the previous moment of the outlier and the battery health state value at the next moment of the outlier;
s32: judging whether the average value of the battery health states of the outliers is larger than a preset battery health state data threshold value;
if yes, determining the outlier as an outlier;
if not, the cluster point is determined as the capacity regeneration point.
According to another specific embodiment of the present invention, in the method for estimating the state of health of a battery of an electric vehicle according to the embodiment of the present invention, the step S3 of averaging the abnormal points includes:
s33: acquiring a battery health state value at the previous moment of the abnormal point and a battery health state value at the next moment of the abnormal point, and calculating an abnormal point average value of the battery health state value at the previous moment of the abnormal point and the battery health state value at the next moment of the abnormal point;
s34: and replacing the abnormal point with an abnormal point mean value of the battery state of health value.
By adopting the scheme, the output data of the regression model is divided into outliers and normal points, the outliers are divided into capacity regeneration points and abnormal points according to the preset battery health state data threshold value by performing mean value calculation on the battery health state value, finally, the abnormal points are subjected to mean value processing, and the obtained capacity regeneration points and the processed abnormal points are substituted into the first prediction model to generate a capacity regeneration result. Therefore, the influence of the capacity regeneration point on the prediction result is fully considered, and the abnormal point is processed, so that the interference of the abnormal point data on the final result is reduced.
According to another specific embodiment of the present invention, the method for estimating the state of health of a battery of an electric vehicle according to the embodiment of the present invention, in step S4, the data distribution balancing processing of the normal point includes:
s41: dividing the data distribution of the normal point into a plurality of areas;
s42: judging whether the ratio of the number of the normal points in any two regions is within a preset ratio range or not;
if yes, reserving normal points of the area;
and if not, repeatedly sampling the area with the smaller number of the normal points.
According to another embodiment of the present invention, the battery state of health estimation method for an electric vehicle disclosed in the embodiment of the present invention, the plurality of regions includes at least three regions, and the preset ratio ranges from 0.5 to 1.5.
By adopting the scheme, the normal point data distribution is divided into a plurality of areas, and the areas with small number of the normal point data are repeatedly sampled, so that the problem of unbalanced data distribution is solved.
The invention has the beneficial effects that:
the method for estimating the battery health state of the electric automobile comprises the steps of firstly acquiring the battery health state parameters of the batteries with the same type in a big data platform, acquiring a database, establishing a regression model according to the battery health state parameter data in the database, forming a partition line by adjusting a regression model constant, dividing output data of the regression model into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset battery health state data threshold, and carrying out mean value processing on the abnormal points, substituting the capacity regeneration points and the processed abnormal points into the first prediction model to generate a capacity regeneration result, carrying out data distribution balance processing on the normal points, substituting the normal points into the second prediction model to generate a normal result, finally replacing the normal result at the same moment with the capacity regeneration result to generate a combined result, and outputting the combined result as an estimation result of the health of the battery. Therefore, the scheme only needs to collect data in the database to establish a regression model, classifies output data according to the model constant, processes the classified data according to the preset threshold and different prediction models, and finally fits the estimation result of the battery health state with higher precision. The whole method flow fully considers the influence of the capacity regeneration phenomenon and the data distribution imbalance on model prediction, and avoids the influence of chemical reaction on battery capacity estimation, so that the estimation result of the battery health state is more accurate.
Drawings
FIG. 1 is a schematic flow chart illustrating a method for estimating a state of health of a battery of an electric vehicle according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of data distribution of normal points provided by an embodiment of the present invention;
FIG. 3 is a diagram illustrating a result of estimating a state of health of a battery according to an embodiment of the present invention;
fig. 4 is a diagram illustrating the estimation result of the state of health of the battery in the related art.
Detailed Description
The following description of the embodiments of the present invention is provided for illustrative purposes, and other advantages and effects of the present invention will become apparent to those skilled in the art from the present disclosure. While the invention will be described in conjunction with the preferred embodiments, it is not intended that features of the invention be limited to these embodiments. On the contrary, the invention is described in connection with the embodiments for the purpose of covering alternatives or modifications that may be extended based on the claims of the present invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The invention may be practiced without these particulars. Moreover, some of the specific details have been left out of the description in order to avoid obscuring or obscuring the focus of the present invention. It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
It should be noted that in this specification, like reference numerals and letters refer to like items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.
In the description of the present embodiment, it should be noted that the terms "upper", "lower", "inner", "bottom", and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings or orientations or positional relationships that are conventionally placed when the products of the present invention are used, and are only used for convenience in describing the present invention and simplifying the description, but do not indicate or imply that the devices or elements indicated must have specific orientations, be configured in specific orientations, and operate, and thus, should not be construed as limiting the present invention.
The terms "first," "second," and the like are used solely to distinguish one from another and are not to be construed as indicating or implying relative importance.
In the description of the present embodiment, it should be further noted that, unless explicitly stated or limited otherwise, the terms "disposed," "connected," and "connected" are to be interpreted broadly, e.g., as a fixed connection, a detachable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present embodiment can be understood in specific cases by those of ordinary skill in the art.
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
The method aims to solve the problem that the influence of a capacity regeneration phenomenon on the health state of the battery is not considered in the prior art, so that the final prediction accuracy of the health state of the battery is not high. The embodiment of the invention discloses a method for estimating the battery health state of an electric automobile, and with reference to fig. 1, the method for estimating the battery health state of the electric automobile disclosed by the embodiment of the invention specifically comprises the following steps:
s1: acquiring battery health state parameters of batteries of the same type by using a big data platform, acquiring a database of the battery health state parameters, and establishing a regression model according to the battery health state parameter data in the database;
s2: analyzing the battery health state parameters by using a regression model to obtain output data of the regression model; the output data of the regression model comprises corrected battery health state information of the electric automobile;
s3: dividing output data into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset battery health state data threshold, and carrying out mean value processing on the abnormal points; substituting the capacity regeneration point and the processed abnormal point average value into a first prediction model to generate a capacity regeneration result;
s4: carrying out data distribution balance processing on the normal points, and substituting the processed normal points into a second prediction model to generate a normal result;
s5: and replacing the normal result at the same moment with the capacity regeneration result to generate a combined result, and outputting the combined result as an estimation result of the battery health state of the electric automobile.
By adopting the scheme, the battery health state parameters of the batteries with the same model in the big data platform are collected to establish a database of the battery health state parameters, then a regression model is established by utilizing the battery health parameter data in the database, and the data in the model is analyzed and trained to divide the output data of the regression model into outliers and normal points. And then classifying the outliers into capacity regeneration points and abnormal points, substituting the capacity regeneration points and the abnormal points subjected to mean value processing into a first prediction model, performing data distribution processing on the normal points, substituting the normal points into a second prediction model, and finally generating a combination result and outputting the combination result as an estimation result of the battery health state of the electric automobile. In the whole process of the method, no matter the battery health state parameters are collected, the regression model is built, or the output data of the regression model is analyzed and processed to generate a combined result, the method can be realized by building a mathematical model or software, actual measurement is not needed, and the operation is convenient. Moreover, the influence of the capacity regeneration phenomenon is comprehensively considered when the battery health state is estimated, and the influence of chemical reaction on the battery capacity estimation is avoided, so that the prediction accuracy of the battery health state is higher.
Next, a method for estimating a state of health of a battery of an electric vehicle according to an embodiment of the present invention will be described in detail with reference to fig. 1 to 4.
Firstly, step S1 is executed, the big data platform is used to collect the battery health status parameters of the batteries with the same model, a database of the battery health status parameters is obtained, and a regression model is established according to the battery health status parameter data in the database.
In this embodiment, the big data platform stores battery health status information of a plurality of batteries, wherein the plurality of batteries may include a plurality of types of batteries or only one type of battery. And the battery state of health information of each battery is a piece of standard data in a big data platform. Firstly, the battery health state information of the batteries with the same model is selected and collected from a big data platform, and then a database is established by the battery health state information of the batteries with the same model. And then training the battery health state information in the database to establish a regression model.
It should be noted that, in this embodiment, the big data platform adopts a Hadoop architecture. Of course, those skilled in the art can select a big data platform with other architectures, which is not limited in this embodiment. The big data platform of the Hadoop system architecture is mainly divided into five layers, wherein the bottommost layer is an acquisition layer and is responsible for acquiring platform data. The data sources of the big data platform comprise the data of the vehicle-mounted terminal, the log stream and the third-party platform. Above the acquisition layer is a big data layer. And the big data layer carries out cluster classification on the data collected by the collection layer. The data of the acquisition layer firstly enters a high-speed service bus of a big data layer, then the big data layer calculates the data in real time and stores the data into a cache cluster or stores the data into a standard data HDFS (Hadoop distributed file system) cluster, an index data cluster and a relational data cluster through a uniform interface of the data layer, and the HDFS has a data storage function. And an analysis layer is arranged above the big data layer and can analyze and calculate the data transmitted by the big data layer. The analysis layer has multiple engines of real-time calculation, off-line calculation, graph calculation, machine learning, situation perception, etc., has the capability of performing business rule modeling, label rule modeling, cleaning/sensing structuring/statistic modeling, and can analyze data cleaning and charging behavior, driving capability, driving behavior, fault tracing, vehicle portrayal, etc. Above the analysis layer is a service layer, which can provide multiple services by using the processing result of the analysis layer. The service layer comprises a plurality of cloud platforms, including a user cloud, a monitoring cloud, a fault cloud, an operation and maintenance cloud and an expert decision cloud, and can provide a plurality of functions such as container hosting, mirror image warehouse and service management. The big data platform is also provided with a display layer, and is mainly used for displaying a large screen, displaying WeChat services and displaying APP which are researched and developed through the platform and are respectively suitable for android and 1OS systems.
Specifically, in this embodiment, the battery state of health parameter includes a battery temperature, a battery current, a battery voltage, and initial battery state of health information.
The battery temperature refers to the actual temperature of the battery, the battery current is the total current of the power battery of the electric automobile, and the battery voltage is the total voltage of the power battery of the electric automobile. The initial battery state of health information is correction information of the battery state of health information estimated based on the traditional method. That is, the battery state of health information obtained based on the conventional method has a large error, and modeling according to the information makes modeling complicated, and generally requires revising the battery state of health information. The battery health state information obtained by correcting the battery health state information is the initial battery health state information in this embodiment.
It should be noted that, the establishing of the regression model according to the battery health state parameter data in the database is to divide the battery health state parameter data in the database into a training set and a test set, establish an initial model according to the training set, and verify the initial model through the test set to obtain the regression model.
Specifically, the battery temperature, the battery current, the battery voltage and the initial battery health state information under different working conditions are stored in a big data platform, a database of required battery health state parameter data is obtained according to the prediction precision requirement, the battery temperature, the battery current and the battery voltage are called from the database as the input of a regression model, the initial battery health state information is used as the output of the model, one part of the battery temperature, the battery current, the battery voltage and the initial battery health state information are used as a training set of the regression model to determine the equation coefficients of the model, the other part is used as a test set to verify the model, for example, the battery temperature, the battery current and the battery voltage in the test set are substituted into the regression model, a corrected battery health state information is calculated, the initial battery health state information in the test set is used for comparison, and verifying whether the obtained corrected battery health state information is accurate.
It should be further noted that there are various ways to divide the data set, and this embodiment is not particularly limited to this.
Next, step S2 is executed to analyze the battery state of health parameter by using the regression model, and obtain the output data of the regression model.
In this embodiment, the output data of the regression model includes corrected battery health status information of the electric vehicle.
The correction of the battery state of health information refers to the battery state of health information that is output after the initial battery state of health information and other battery state of health parameters are input to a regression model and then analyzed and calculated by the regression model.
Preferably, in the method for estimating the state of health of a battery of an electric vehicle according to the embodiment of the present invention, the regression model is a linear regression model (SVR regression model). Of course, those skilled in the art can select other regression models, which is not limited in this embodiment.
Specifically, in this embodiment, the process of analyzing the battery state of health parameter by using the regression model to obtain the output data of the regression model specifically includes: and fitting a straight line which accords with most of data sets by using the input data and the output data of the regression model, and when new battery temperature, battery current and battery voltage are input, correcting the health state information of the battery on the straight line or near the straight line according to the regression model.
Next, step S3 is executed to divide the output data into outliers and normal points, divide the outliers into capacity regeneration points and abnormal points according to the preset threshold of the battery health status data, and perform an average processing on the abnormal points; and substituting the capacity regeneration point and the processed abnormal point average value into the first prediction model to generate a capacity regeneration result.
Specifically, in step S3, the distinguishing of the output data into outliers and normal points includes: and obtaining model parameters of the regression model. The model parameters comprise independent variables, dependent variables and constants; and obtaining a dividing line according to the independent variable, the dependent variable and the adjusted constant by adjusting the value of the constant, and dividing the output data of the regression model into outliers and normal points by using the dividing line.
Specifically, taking the outlier recognition model as the unary linear model y ═ kx + b as an example, two parallel lines can be obtained by adjusting b, and at this time, the values other than the parallel lines are considered as outliers, and the points on and in the lines are both normal points.
Further, in step S3, the dividing the outlier into a capacity regeneration point and an abnormal point according to the preset threshold of the battery state of health data includes:
s31: acquiring a battery health state value at the previous moment of each outlier and a battery health state value at the next moment of the outlier, and calculating a battery health state mean value of the outlier according to the battery health state value at the previous moment of the outlier and the battery health state value at the next moment of the outlier;
s32: judging whether the average value of the battery health states of the outliers is larger than a preset battery health state data threshold value;
if yes, determining the outlier as an outlier;
if not, the cluster point is determined as the capacity regeneration point.
Specifically, if the current state of health mean value is greater than the preset battery state of health data threshold, it is a point far from the data point, and it is determined as an abnormal point, and conversely, it is determined as a capacity regeneration point.
Note that the preset battery state of health data threshold value is a value that separates a capacity regeneration outlier from an abnormal point at the time of data collection. The preset battery state of health data threshold may be obtained in two ways: firstly, a threshold value can be used as a hyper-parameter, and the fitting is automatically identified through an algorithm; the second is a statistical empirical method, in which the value q of y ═ kx + b is used as a standard, for example, (q +1/4q, q-1/4q) is used as an upper and lower bound threshold. Those skilled in the art can select the method according to actual needs, and the embodiment is not limited in this respect.
Specifically, a new b value is adjusted to obtain two new parallel lines, at this time, an average value is calculated for points other than the parallel lines, that is, an average value of the battery health state value at the previous time and the battery health state value at the next time is calculated, if the current battery health state average value is outside the parallel line at this time, the point is determined as an abnormal point, and if the current battery health state average value is within the parallel line, the point is determined as a capacity regeneration point.
Further, in step S3, the average processing is performed on the abnormal point, which includes:
s33: acquiring a battery health state value at the previous moment of the abnormal point and a battery health state value at the next moment of the abnormal point, and calculating an abnormal point average value of the battery health state value at the previous moment of the abnormal point and the battery health state value at the next moment of the abnormal point;
s34: and replacing the abnormal point with an abnormal point mean value of the battery state of health value.
Specifically, since the abnormal point has a great influence on the model prediction, the abnormal point is determined by obtaining the battery state of health value at the previous time and the battery state of health value at the subsequent time, and calculating the battery state of health at the previous timeAnd replacing the abnormal value at the current moment with the obtained average value to predict the model. For example, the state of health value of the battery at time t-1 is SOHt-189, and the state of health value of the battery is SOH at time ttWhen the battery state of health value is SOH at time t +1, 97t+1And (90), obtaining the abnormal value of the value at the time t through an outlier identification model, discarding the abnormal value, replacing the abnormal value by using a mean value due to the fact that the battery state of health value has high time dependence and does not suddenly change, and obtaining the SOH through the formulat＝(SOHt-1+SOHt+1) And/2, so the mean value at time t is 89.5.
Preferably, the first prediction model is an extreme gradient boost model (Xgboost model). Of course, those skilled in the art can select other prediction models as needed, and the embodiment is not limited thereto.
It should be noted that the capacity regeneration point and the anomaly point processed by the average are used as input of the first prediction model, which is an algorithm model that trains historical data to predict the trend of the value after the first prediction model. Through training of a large amount of data in the database, the first prediction model has the capability of predicting the trend, when the sensor collects data and inputs the data, the position where an outlier of the subsequent capacity regeneration appears can be predicted, and a capacity regeneration result is generated through the first prediction model.
Next, step S4 is executed to perform data distribution balancing processing on the normal points, and to substitute the processed normal points into the second prediction model to generate a normal result.
Further, in step S4, the data distribution balancing process is performed on the normal point, and includes:
s41: dividing the data distribution of the normal point into a plurality of areas;
s42: judging whether the ratio of the number of the normal points in any two regions is within a preset ratio range or not;
if yes, reserving normal points of the area;
and if not, repeatedly sampling the area with the smaller number of the normal points.
It should be noted that, in this embodiment, the plurality of regions includes at least three regions, and the preset ratio ranges from 0.5 to 1.5. The preset ratio range may be 0.5, 0.7, 0.9, 1.1, 1.3, 1.5, or other values in the range, which is not limited in this embodiment.
Specifically, the data distribution of the normal point may include three regions, four regions, or even more regions, which is not limited by the embodiment.
Specifically, referring to the schematic diagram of the unbalanced distribution of the normal point data provided by the embodiment of the present invention shown in fig. 4, it can be seen that the data distribution of the normal point is divided into three regions, i.e., a small current region, a medium current region, and a large current region. In the current as input, it can be found from the distribution that the sample data is mainly accumulated in a small current or medium current range, and the data in a large current area is less, so that the data distribution balance processing is required.
Preferably, in this embodiment, the data distribution balancing process may be an F divergence check. The distribution of samples is detected, samples with less distribution (such as samples in a large current area) are repeatedly sampled during calculation, the number of the samples is increased, and the imbalance of the distribution of the samples is improved.
Preferably, the second prediction model is any one of a self-attention model (Transformer model), a long-short term memory network (LSTM model), a gated cyclic unit network (GRU model), and a recurrent neural network (RNN model), which is not particularly limited in this embodiment.
Furthermore, after data distribution equalization processing is carried out on the normal point data, all the normal points are used as input of a second prediction model, the second prediction model also has the capability of predicting trends, and when data collected by a sensor are input, the positions of the subsequent normal points can be predicted.
Next, step S5 is executed to replace the normal result at the same time with the capacity regeneration result, generate a combination result, and output the combination result as an estimation result of the battery state of health of the electric vehicle.
Specifically, when current, voltage and temperature parameters are input into a first prediction model trained by historical data, new capacity regeneration data can be obtained by directly calling the first prediction model, similarly, a data curve without an outlier can be obtained by inputting the data curve into a second prediction model, the capacity regeneration data and the outlier are fused, and the capacity regeneration data is larger than the predicted value of the second prediction model at the same position due to capacity regeneration, and at the moment, the predicted value of the second prediction model at the same position is replaced by the capacity regeneration data predicted by the first prediction model. The final prediction curve obtained at this time is the estimation result of the state of health of the battery.
In this embodiment, referring to a schematic diagram of an estimation result of a battery health state provided in an embodiment of the present invention shown in fig. 3, an outlier recognition model is used to divide regression model output data into outliers and normal points, divide the outliers into capacity regeneration points and abnormal points according to a preset threshold, perform mean processing on the abnormal points, use an average of the capacity regeneration points and the processed abnormal points as an input of a first prediction model and output a capacity regeneration result, perform data distribution imbalance processing on the normal points, use the processed data as an input of a second prediction model and generate a normal result, replace the normal result at the same time with the capacity regeneration result, and generate a combined result, so as to obtain the estimation result of the battery health state shown in fig. 3.
Referring to the diagram of the estimation result of the battery state of health in the prior art shown in fig. 4, compared with the estimation result of the battery state of health in fig. 3, the fitting degree of the prediction data and the test data shown in fig. 4 is not high, and the deviation is large. The influence of the capacity regeneration phenomenon on the battery health state is fully considered in the prediction result shown in fig. 3, and the abnormal data of the model is processed, so that the obtained prediction result has high fitting rate on the test data, and the result is more accurate.
By adopting the scheme, firstly, the battery health state parameters of the batteries with the same model in the big data platform are collected to obtain the database, a regression model is established according to the battery health state parameter data in the database, forming a partition line by adjusting a regression model constant, dividing output data of the regression model into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset battery health state data threshold, and carrying out mean value processing on the abnormal points, substituting the capacity regeneration points and the processed abnormal points into the first prediction model to generate a capacity regeneration result, carrying out data distribution balance processing on the normal points, substituting the normal points into the second prediction model to generate a normal result, finally replacing the normal result at the same moment with the capacity regeneration result to generate a combined result, and outputting the combined result as an estimation result of the health of the battery. Therefore, the scheme only needs to collect data in the database to establish a regression model, classifies output data according to the model constant, processes the classified data according to the preset threshold and different prediction models, and finally fits the estimation result of the battery health state with higher precision. The whole method flow fully considers the influence of the capacity regeneration phenomenon and the data distribution imbalance on model prediction, and avoids the influence of chemical reaction on battery capacity estimation, so that the estimation result of the battery health state is more accurate.
While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that the foregoing is a more detailed description of the invention, taken in conjunction with the specific embodiments thereof, and that no limitation of the invention is intended thereby. Various changes in form and detail, including simple deductions or substitutions, may be made by those skilled in the art without departing from the spirit and scope of the invention.
Claims (9)
1. A method for estimating the state of health of a battery of an electric vehicle is characterized by comprising the following steps:
s1: acquiring battery health state parameters of batteries of the same type by using a big data platform, acquiring a database of the battery health state parameters, and establishing a regression model according to the battery health state parameter data in the database;
s2: analyzing the battery health state parameters by using the regression model to obtain output data of the regression model; wherein the output data of the regression model comprises corrected battery health state information of the electric automobile;
s3: dividing the output data into outliers and normal points, dividing the outliers into capacity regeneration points and abnormal points according to a preset battery health state data threshold, and carrying out mean value processing on the abnormal points; substituting the capacity regeneration point and the processed abnormal point average value into a first prediction model to generate a capacity regeneration result;
s4: carrying out data distribution balance processing on the normal points, and substituting the processed normal points into a second prediction model to generate a normal result;
s5: and replacing the normal result at the same moment with the capacity regeneration result to generate a combined result, and outputting the combined result as an estimation result of the battery health state of the electric automobile.
2. The method of estimating battery state of health of an electric vehicle of claim 1, wherein the battery state of health parameters comprise battery temperature, battery current, battery voltage, initial battery state of health information.
3. The method for estimating the state of health of the battery of the electric vehicle according to claim 2, wherein in step S1, the establishing the regression model according to the battery state of health parameter data in the database comprises:
the regression model is obtained by dividing the battery health state parameter data in the database into a training set and a test set, establishing an initial model according to the training set and verifying the initial model through the test set.
4. The method of estimating the state of health of a battery of an electric vehicle according to claim 3, wherein the regression model is a linear regression model; and the number of the first and second electrodes,
the first prediction model is an extreme gradient lifting model; and the number of the first and second electrodes,
the second prediction model is any one of a self-attention model, a long-short term memory network, a gated cyclic unit network and a cyclic neural network.
5. The method for estimating state of health of battery of electric vehicle according to claim 4, wherein said step S3 of distinguishing said output data into outliers and normals comprises:
obtaining model parameters of the regression model; wherein the model parameters include independent variables, dependent variables, and constants;
and obtaining a dividing line according to the independent variable, the dependent variable and the adjusted constant by adjusting the numerical value of the constant, and distinguishing the output data of the regression model into the outlier and the normal point by using the dividing line.
6. The method for estimating state of health of battery of electric vehicle according to claim 5, wherein said step S3, dividing said outlier into a capacity regeneration point and an abnormal point according to a preset threshold of battery state of health data, comprises:
s31: acquiring a battery health state value at the moment before each outlier and a battery health state value at the moment after each outlier, and calculating a battery health state mean value of each outlier according to the battery health state value at the moment before each outlier and the battery health state value at the moment after each outlier;
s32: judging whether the average value of the battery health states of the outliers is larger than a preset battery health state data threshold value;
if yes, determining the outlier as an abnormal point;
and if not, determining the outlier as a capacity regeneration point.
7. The method for estimating the state of health of the battery of the electric vehicle according to claim 6, wherein the step S3 of averaging the outliers comprises:
s33: acquiring a battery health state value at the previous moment of the abnormal point and a battery health state value at the next moment of the abnormal point, and calculating an abnormal point average value of the battery health state value at the previous moment of the abnormal point and the battery health state value at the next moment of the abnormal point;
s34: and replacing the abnormal point with an abnormal point mean value of the battery state of health value.
8. The method for estimating the state of health of the battery of the electric vehicle according to claim 7, wherein the step S4 of performing data distribution balancing processing on the normal point includes:
s41: dividing the data distribution of the normal point into a plurality of areas;
s42: judging whether the ratio of the number of the normal points in any two regions is within a preset ratio range or not;
if so, reserving normal points of the area;
and if not, repeatedly sampling the area with the smaller number of the normal points.
9. The method of estimating the state of health of a battery of an electric vehicle according to claim 8, wherein the plurality of regions includes at least three of the regions, and the preset ratio ranges from 0.5 to 1.5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011411897.9A CN112505570A (en) | 2020-12-03 | 2020-12-03 | Method for estimating battery health state of electric automobile |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011411897.9A CN112505570A (en) | 2020-12-03 | 2020-12-03 | Method for estimating battery health state of electric automobile |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112505570A true CN112505570A (en) | 2021-03-16 |
Family
ID=74970334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011411897.9A Pending CN112505570A (en) | 2020-12-03 | 2020-12-03 | Method for estimating battery health state of electric automobile |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112505570A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113075574A (en) * | 2021-03-30 | 2021-07-06 | 上海交通大学 | Battery health state prediction method and equipment based on self-adaptive information fusion |
| CN114460481A (en) * | 2022-01-27 | 2022-05-10 | 重庆邮电大学 | Energy storage battery thermal runaway early warning method based on Bi-LSTM and attention mechanism |
| CN115248377A (en) * | 2022-09-21 | 2022-10-28 | 北京科锐特科技有限公司 | Asynchronous motor rotor broken bar fault detection method, computer device and medium |
| CN115389948A (en) * | 2022-10-28 | 2022-11-25 | 昆山斯沃普智能装备有限公司 | Method for establishing health state evaluation model of power battery of battery replacement station |
| US20230294554A1 (en) * | 2022-03-18 | 2023-09-21 | GM Global Technology Operations LLC | Battery health monitoring and failure identification |
| CN116846040A (en) * | 2023-08-31 | 2023-10-03 | 合肥博雷电气有限公司 | All-digital control method for constant-voltage constant-current charge-discharge power supply |
| WO2023221587A1 (en) * | 2022-05-16 | 2023-11-23 | 深圳市道通合创数字能源有限公司 | Method for determining state of health of power battery of electric vehicle, and server |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344923A (en) * | 2013-08-01 | 2013-10-09 | 哈尔滨工业大学 | EKF (extended Kalmar filter)-method and NSDP-AR-model-based method for predicting cycling life of fusion-type lithium ion battery |
| CN106093796A (en) * | 2016-08-09 | 2016-11-09 | 北京航空航天大学 | Based on regression criterion reconstruction in frequency-domain have lithium battery capacity and life-span prediction method in the case of capacity regeneration |
| CN106339755A (en) * | 2016-08-29 | 2017-01-18 | 深圳市计量质量检测研究院 | Lithium battery SOH (State of Health) prediction method based on neural network and periodic kernel functions GPR |
| CN110988694A (en) * | 2019-12-17 | 2020-04-10 | 南京航空航天大学 | Lithium ion battery residual life prediction method and system aiming at capacity regeneration |
| CN111398837A (en) * | 2020-04-01 | 2020-07-10 | 重庆大学 | Vehicle battery health state estimation method based on data driving |
| DE102019007510A1 (en) * | 2019-10-28 | 2020-07-16 | Daimler Ag | Method for determining a state of health of an electrical energy store by means of a stored capacity model depending on an anode overhang effect, and electronic computing device |
-
2020
- 2020-12-03 CN CN202011411897.9A patent/CN112505570A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103344923A (en) * | 2013-08-01 | 2013-10-09 | 哈尔滨工业大学 | EKF (extended Kalmar filter)-method and NSDP-AR-model-based method for predicting cycling life of fusion-type lithium ion battery |
| CN106093796A (en) * | 2016-08-09 | 2016-11-09 | 北京航空航天大学 | Based on regression criterion reconstruction in frequency-domain have lithium battery capacity and life-span prediction method in the case of capacity regeneration |
| CN106339755A (en) * | 2016-08-29 | 2017-01-18 | 深圳市计量质量检测研究院 | Lithium battery SOH (State of Health) prediction method based on neural network and periodic kernel functions GPR |
| DE102019007510A1 (en) * | 2019-10-28 | 2020-07-16 | Daimler Ag | Method for determining a state of health of an electrical energy store by means of a stored capacity model depending on an anode overhang effect, and electronic computing device |
| CN110988694A (en) * | 2019-12-17 | 2020-04-10 | 南京航空航天大学 | Lithium ion battery residual life prediction method and system aiming at capacity regeneration |
| CN111398837A (en) * | 2020-04-01 | 2020-07-10 | 重庆大学 | Vehicle battery health state estimation method based on data driving |
Non-Patent Citations (1)
| Title |
|---|
| 王友仁等: "基于变分模态分解和高斯过程回归的锂离子电池剩余寿命预测方法", 计算机与现代化, no. 02, pages 83 - 88 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113075574A (en) * | 2021-03-30 | 2021-07-06 | 上海交通大学 | Battery health state prediction method and equipment based on self-adaptive information fusion |
| CN113075574B (en) * | 2021-03-30 | 2022-09-09 | 上海交通大学 | Battery health state prediction method and equipment based on self-adaptive information fusion |
| CN114460481A (en) * | 2022-01-27 | 2022-05-10 | 重庆邮电大学 | Energy storage battery thermal runaway early warning method based on Bi-LSTM and attention mechanism |
| CN114460481B (en) * | 2022-01-27 | 2024-09-03 | 浙江商储数智能源科技有限公司 | Bi-LSTM and attention mechanism-based energy storage battery thermal runaway early warning method |
| US20230294554A1 (en) * | 2022-03-18 | 2023-09-21 | GM Global Technology Operations LLC | Battery health monitoring and failure identification |
| US11981228B2 (en) * | 2022-03-18 | 2024-05-14 | GM Global Technology Operations LLC | Battery health monitoring and failure identification |
| WO2023221587A1 (en) * | 2022-05-16 | 2023-11-23 | 深圳市道通合创数字能源有限公司 | Method for determining state of health of power battery of electric vehicle, and server |
| CN115248377A (en) * | 2022-09-21 | 2022-10-28 | 北京科锐特科技有限公司 | Asynchronous motor rotor broken bar fault detection method, computer device and medium |
| CN115248377B (en) * | 2022-09-21 | 2022-12-13 | 北京科锐特科技有限公司 | Asynchronous motor rotor broken bar fault detection method, computer equipment and medium |
| CN115389948A (en) * | 2022-10-28 | 2022-11-25 | 昆山斯沃普智能装备有限公司 | Method for establishing health state evaluation model of power battery of battery replacement station |
| CN116846040A (en) * | 2023-08-31 | 2023-10-03 | 合肥博雷电气有限公司 | All-digital control method for constant-voltage constant-current charge-discharge power supply |
| CN116846040B (en) * | 2023-08-31 | 2023-11-14 | 合肥博雷电气有限公司 | All-digital control method for constant-voltage constant-current charge-discharge power supply |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112505570A (en) | Method for estimating battery health state of electric automobile | |
| Guo et al. | Online estimation of SOH for lithium-ion battery based on SSA-Elman neural network | |
| CN112904219B (en) | Big data-based power battery health state prediction method | |
| CN112765772A (en) | Power battery residual life prediction method based on data driving | |
| CN109947086A (en) | Mechanical breakdown migration diagnostic method and system based on confrontation study | |
| CN111241154A (en) | Storage battery fault early warning method and system based on big data | |
| CN109143074B (en) | Power battery model parameter identification method and system | |
| US20230305073A1 (en) | Method and apparatus for providing a predicted aging state of a device battery based on a predicted usage pattern | |
| CN111191824B (en) | Power battery capacity attenuation prediction method and system | |
| CN117783745B (en) | Data online monitoring method and system for battery replacement cabinet | |
| CN116609686B (en) | Battery cell consistency assessment method based on cloud platform big data | |
| CN111157908A (en) | Battery health degree prediction method based on big data machine learning | |
| CN114036647A (en) | Power battery safety risk assessment method based on real vehicle data | |
| CN114167284A (en) | Lithium battery RUL prediction method and device based on BMS big data and ensemble learning | |
| CN117406027A (en) | Distribution network fault distance measurement method and system | |
| CN111144021A (en) | Fuel cell service life prediction method and system | |
| CN113424119A (en) | Work efficiency evaluation method, work efficiency evaluation device, and program | |
| CN112070535A (en) | Electric vehicle price prediction method and device | |
| CN118050653A (en) | Method and device for predicting the battery of a diagnostic device using a multi-element converter model | |
| Li et al. | State of health estimation and prediction of electric vehicle power battery based on operational vehicle data | |
| CN110988693A (en) | Chemical battery energy storage application analysis method based on modeling simulation | |
| CN110596595A (en) | Method for predicting RUL of rail-traffic lithium battery through big data | |
| CN112816788B (en) | Capacity value estimation method of locomotive traction converter direct current support capacitor and application system thereof | |
| CN114840983A (en) | Power storage battery SOH calculation model construction method and calculation method | |
| CN113962558A (en) | Industrial internet platform evaluation method and system based on production data management |
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 | ||
| AD01 | Patent right deemed abandoned | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20240927 |