CN108490365A - A method of the remaining life of the power battery of estimation electric vehicle - Google Patents

A method of the remaining life of the power battery of estimation electric vehicle Download PDF

Info

Publication number
CN108490365A
CN108490365A CN201810349151.6A CN201810349151A CN108490365A CN 108490365 A CN108490365 A CN 108490365A CN 201810349151 A CN201810349151 A CN 201810349151A CN 108490365 A CN108490365 A CN 108490365A
Authority
CN
China
Prior art keywords
power battery
linear
value
identified
soh
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.)
Granted
Application number
CN201810349151.6A
Other languages
Chinese (zh)
Other versions
CN108490365B (en
Inventor
熊瑞
张永志
何洪文
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN201810349151.6A priority Critical patent/CN108490365B/en
Publication of CN108490365A publication Critical patent/CN108490365A/en
Application granted granted Critical
Publication of CN108490365B publication Critical patent/CN108490365B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Secondary Cells (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)
  • Tests Of Electric Status Of Batteries (AREA)

Abstract

The present invention relates to the methods that the remaining life of the power battery management system of electric vehicle more particularly to a kind of power battery to electric vehicle is estimated.Estimate that precision existing for the remaining life of power battery is low to solve the prior art, problem of high cost, the present invention proposes that a kind of method of the power battery remaining life of estimation electric vehicle, On-line Estimation go out the real-time remaining capacity value C of power batterym;The real time health state SOH of power battery is calculated,Determine the linear attenuation stage starting point of power battery;Establish the linear Ageing Model SOH of power batterykAnd the parameter of cutting edge aligned Ageing Model is recognized, the remaining charge and discharge cycles number p of estimated driving force battery,This method of estimation obtains the remaining life of power battery by On-line Estimation, simple and convenient, and estimated accuracy is high, at low cost.

Description

A method of the remaining life of the power battery of estimation electric vehicle
Technical field
The present invention relates to the power battery management system of electric vehicle more particularly to a kind of power batteries to electric vehicle The method estimated of remaining life.
Background technology
Since lithium-ion-power cell has the characteristics such as high-energy/power density and long life cycle, therefore it is widely used In electric vehicle field.The performance of power battery constantly decays during the work time, when its health status value decays to healthy shape State initial value 80% when, then it is assumed that power battery reach cut-off the service life (End of life, abbreviation EOL).It can be seen that carrying Preceding remaining life (Remaining useful life, abbreviation to the power battery in power battery especially electric vehicle RUL it) is estimated, user can be helped to repair maintenance to power battery in advance, avoid leading to electricity because power battery fails Electrical automobile can not use, or even cause traffic accident, and personnel and property loss are caused to user.
Currently, the method estimated the remaining life of power battery can be divided into the method for estimation based on model and be based on Two class of method of estimation of data-driven.Wherein, the method for estimation based on model is to combine filtering skill using non-linear Ageing Model Art estimates the remaining life of power battery, and this method needs first according to the test experiment data of power battery to non-linear old Change model to be initialized, reuses filtering technique and non-linear Ageing Model is carried out according to the part online data of power battery Estimating to the remaining life of power battery is completed in update.But since the driving cycle of electric vehicle is complicated and changeable, thus it is real The personnel of testing are difficult to design to work online the identical accelerated life test of condition with the power battery of electric vehicle to acquire Effective off-line data, and then accurate prediction result can not be obtained.
Predictor method based on data-driven is based on offline or on-line training data, using machine learning techniques to dynamic The remaining life of power battery is estimated, be applied to now the machine learning techniques that power battery remaining life is estimated have it is adaptive Answer Recognition with Recurrent Neural Network (Adaptive recurrent neural network, abbreviation ARNN) technology, support vector machines (Support vector machine, abbreviation SVM) technology and Method Using Relevance Vector Machine (Relevance vector machine, Abbreviation RVM) technology.Machine learning techniques be divided into for based on off-line data machine learning method and based on the machine of online data Two class of learning method.Wherein, the machine learning method based on off-line data needs a large amount of off-line data that could realize to power Effective training of battery model, and then the remaining life for obtaining power battery is estimated, operand is big, and estimated cost is high.Based on The machine learning method of line number evidence does not need off-line data, is completely dependent on online data and realizes to the remaining life of power battery It estimates, and estimates be required to constantly train all online datas each time, but this kind of method does not account for completely The attenuation characteristic of power battery aging, therefore the remaining life accuracy for the power battery estimated is low, it is unreliable.
Invention content
Estimate that precision existing for the remaining life of power battery is low to solve the prior art, problem of high cost, the present invention It is proposed a kind of method of the power battery remaining life of estimation electric vehicle, this method comprises the following steps:
Step S1, in the power battery course of work, charging and discharging currents and the end of the power battery are acquired in real time Voltage, and On-line Estimation goes out the real-time remaining capacity value C of the power batterym
Step S2, according to the real-time remaining capacity value C of the power batterymWith the initial capacity value C of the power battery0, The real time health state SOH of the power battery is calculated, and
Step S3, turning to the decaying track of the real time health state value SOH of the power battery and the power battery The corresponding health status value SOH of point0It is compared, when the real time health state value SOH of the power battery is less than the power The corresponding health status SOH of decaying track inflection point of battery0When, the real time health state value SOH of the power battery0It is corresponding Charge and discharge cycles number k0For the starting point in the linear attenuation stage of the power battery;
Step S4, the linear of the power battery is established according to the decaying track in the linear attenuation stage of the power battery Ageing Model:SOHk=b0+b1·k+εk, εk~N (0, σ2),
Wherein,
K indicates the power battery in current time corresponding charge and discharge cycles number,
b0, b1Indicate the parameter to be identified of the linear Ageing Model of the power battery,
εkIt is 0 for mean value, variance σ2Independent normal stochastic error,
N indicates normal distribution,
And according to the charging and discharging currents and terminal voltage value of the power battery acquired in real time, on-line identification obtains the line The parameter to be identified of property Ageing Model;
Step S5, estimate the remaining charge and discharge cycles number of the power battery, and
Wherein,
The power battery is indicated in cut-off time in service life corresponding health status value,
L indicates that the power battery lights the charge and discharge cycles time carried out to current time from the starting of linear attenuation stage Number,
P indicates the power battery from current time to the remaining charge and discharge cycles number of cut-off time in service life.It is this Estimate that the method for the remaining life of the power battery of electric vehicle carries out the online data of power battery by way of recursion Interative computation accurately estimates out the remaining life of power battery, effectively promotes the stability and reliability of estimation results.Separately Outside, this method of estimation is both not required to a large amount of off-line datas of acquisition power battery, is also not required to instruct online data repeatedly Practice, it is simple and convenient, greatly reduce the cost for the remaining life for estimating power battery.
Pair preferably, in the step S3, accelerated aging is carried out with the power battery of the power battery same model Experiment, obtains the starting point in the linear attenuation stage of the power battery.
Preferably, it in the step S4, is picked out using the least square method of recursion with forgetting factor described linear The parameter to be identified of Ageing Model.
Preferably, identification process is as follows:
First, Recursive Least Squares Estimation device of the initialization with forgetting factor,
Wherein,
θ indicates the parameter vector to be identified of the linear Ageing Model of the power battery, and θ=[b0, b1]T, T expression squares Battle array transposition,
For the initial estimate of the parameter vector to be identified of the linear Ageing Model of the power battery, setting
E expressions take average operation,
Q indicates the error covariance square of the parameter vector estimated value to be identified of the linear Ageing Model of the power battery Battle array,
Q0For the error co-variance matrix of the parameter vector estimated value to be identified of the linear Ageing Model of the power battery Initial value, set Q0=106I, and I is unit matrix;
Then, the power battery is inputted into the Recursive Least Squares Estimation device with forgetting factor linearly declining The number i of derogatory section of linear attenuation data point, and i=1,2,3, L, the linear attenuation data point includes the power electric The charge and discharge cycles number k and the power battery in pond corresponding health status value SOH when charge and discharge cycles number is kk,
Update gain matrix Ki, and
Update the parameter vector estimated value to be identified of the linear Ageing Model of the power batteryAnd
Update the error co-variance matrix of the parameter vector estimated value to be identified of the linear Ageing Model of the power battery Qi, and
Wherein,
Indicate the input vector of the Recursive Least Squares Estimation device with forgetting factor, and
Y indicates the observation of the Recursive Least Squares Estimation device with forgetting factor, and y=SOHk,
μ indicates forgetting factor, and the value range of μ is (0,1);
Finally, according toPick out the ginseng to be identified of the linear Ageing Model of the power battery Number.
Preferably, the value range of the forgetting factor μ is (0.95,1), in this way, it is ensured that passing with forgetting factor Push away the robust property of LS estimator.
In addition, the present invention also proposes a kind of remaining life of power battery that estimating electric vehicle using any one of the above Method power battery management system that the remaining life of the power battery of electric vehicle is estimated.Such power battery Management system is when the remaining life of the power battery to electric vehicle is estimated, the power battery pipe of relatively existing mainstream Reason system, estimated accuracy is high, safe and reliable and at low cost.
Description of the drawings
Fig. 1 is to carry out accelerated aging reality to four power batteries that cycle discharge-rate is respectively 0.5C, 1C, 1.5C, 2C The decaying track schematic diagram of the health status value SOH for the power battery tested, wherein Fig. 1 (a) is to be to cycle discharge-rate The power battery of 0.5C carries out the decaying track schematic diagram of the health status value SOH for the power battery that accelerated life test obtains, Fig. 1 (b) is the health status that the power battery for being 1C to cycle discharge-rate carries out the power battery that accelerated life test obtains The decaying track schematic diagram of value SOH, Fig. 1 (c) are that the power battery for being 1.5C to cycle discharge-rate carries out accelerated life test The decaying track schematic diagram of the health status value SOH of obtained power battery, it is that 2C is moved that Fig. 1 (d), which is to cycle discharge-rate, Power battery carries out the decaying track schematic diagram of the health status value SOH for the power battery that accelerated life test obtains;
Fig. 2 is to use method of estimation of the present invention successively to cycle discharge-rate for four power of 0.5C, 1C, 1.5C, 2C The estimated result of the remaining life of battery, wherein Fig. 2 (a) is the remaining longevity for the power battery for being 0.5C to cycle discharge-rate The estimated result of life, Fig. 2 (b) are the estimated result of the remaining life for the power battery for being 1C to cycle discharge-rate, Fig. 2 (c) For the estimated result of the remaining life for the power battery for being 1.5C to cycle discharge-rate, Fig. 2 (d) is to be to cycle discharge-rate The estimated result of the remaining life of the power battery of 2C.
Specific implementation mode
In the following, in conjunction with Fig. 1 and 2, the method for the remaining life of the power battery of present invention estimation electric vehicle is carried out detailed It describes in detail bright.Specific estimation procedure is as follows:
It is real using the power battery management system (BMS) of electric vehicle in the power battery course of work of electric vehicle When acquire the charging and discharging currents and terminal voltage of power battery, and using wearing Vernam model combination H infinity filtering method On-line Estimation Go out the real-time remaining capacity value C of power batterym.Certainly, dual polarization model or the filtering of multistage RC network models coupling can also be used Real-time remaining capacity value C of the method to power batterymCarry out On-line Estimation.
According to the real-time remaining capacity value C of power batterymWith initial capacity value C0, calculate the real time health of power battery State value SOH, and
Since the maximum available of power battery can constantly decay with the progress of charge and discharge, therefore power battery is most Big active volume is continually changing with the increase of charge and discharge cycles number.The process of charge and discharge is carried out in power battery In, when the accumulated discharge capacity of power battery is equal to the rated capacity of the power battery, then it is assumed that the power battery carries out One charge and discharge cycles.It can be seen that after power battery often carries out a charge and discharge cycles, corresponding maximum available is all It can be there are one new value.That is, each cycle-index of power battery corresponds to a maximum available value.Separately Outside, the corresponding new maximum available value of each cycle-index of power battery can be obtained by online data real-time estimation Go out.
During power battery carries out charge and discharge, since the maximum available of power battery can be with its charge and discharge Progress and constantly decay, therefore the health status value SOH of the power battery also can with the progress of the charge and discharge of power battery and Constantly decaying, and then the health status value SOH that power battery can be obtained constantly decays with the increase of charge and discharge cycles number Decaying track.For including four power batteries that 0.5C, 1C, 1.5C, 2C recycle discharge-rate successively, added respectively Fast life experiment can obtain the decaying track of the health status value SOH of four power batteries respectively such as Fig. 1 (a), 1 (b), 1 (c) and shown in 1 (d), therefore, the decaying track of the health status value SOH of power battery is generally divided into multiple stages, including the initial segment Decaying quick the nonlinear attenuation stage and second stage the decaying slow linear attenuation stage, and second stage is linear The tie point in decling phase and the nonlinear attenuation stage of initial period is known as the inflection point of decaying track, the corresponding data of the inflection point Point includes a charge and discharge cycles number and health status value corresponding with the charge and discharge cycles number, and power battery can exist Corresponding charge and discharge cycles number is denoted as k when the inflection point of its track of decaying0, corresponding health status value is denoted as SOH0.Namely It says, the inflection point of the decaying track of power battery had both been known as the terminal in the nonlinear attenuation stage of the power battery, also might be used To be known as the starting point in the linear attenuation stage of the power battery.In addition, passing through the health of the power battery to same model It is observed the decaying track of state value, it is possible to find, for the power battery of a certain model, the inflection point of decaying track is logical An often corresponding specific health status value, the i.e. power battery of the model corresponding healthy shape in the inflection point of its track of decaying State value SOH0It is a particular value.It can be seen that accelerated life test can be carried out by the power battery to flow-through number, it is dynamic to determine The corresponding health status value SOH of inflection point of the decaying track of power battery0, and the remaining longevity of the power battery in estimation electric vehicle When life, by the corresponding health status value SOH of the inflection point of the power battery of the model0As the inflection point for judging the model power battery Preset value, i.e., when the real time health state value SOH of power battery is less than the corresponding health status of inflection point of its track of decaying Value SOH0When, the decaying track of the health status value of the power battery enters the linear attenuation stage, while can be by the power battery The corresponding charge and discharge cycles number k at the inflection point of the decaying track of its health status value0The linear of the power battery is denoted as to decline Derogatory section of starting point.
To sum up, the linear attenuation for obtaining power battery can be recognized according to the real time health state value SOH of power battery Stage starting point, i.e., when the inflection point that the real time health state value SOH of power battery is less than the decaying track of power battery is corresponding Health status value SOH0When, the corresponding charge and discharge cycles number k of power battery0The as linear attenuation stage starting of power battery Point.
The linear Ageing Model of power battery to be estimated is established according to the linear attenuation track of the power battery of the model: SOHk=b0+b1·k+εk, εk~N (0, σ2),
Wherein,
K indicates power battery in current time corresponding charge and discharge cycles number,
b0, b1Indicate the parameter to be identified of the linear Ageing Model of power battery,
εkIndicate mean value be 0, variance σ2Independent normal stochastic error,
N indicates normal distribution.
According to the charging and discharging currents and terminal voltage value of real-time collected power battery, on-line identification obtains power battery The parameter to be identified of linear Ageing Model.
It is constantly to obtain new data with genic least square method relative to common least square method of recursion On the basis of constantly abandon legacy data, the health of power battery can accurately be described by constructing so as to preferably tracking new data The Ageing Model of the decaying track of state value.Therefore, it is preferred to use the least square method of recursion with forgetting factor recognizes power electric The parameter to be identified of the linear Ageing Model in pond.
When carrying out parameter identification, it usually needs inputted into the linear Ageing Model of the power battery of structure a certain amount of Input data could obtain more accurate model parameter.Therefore, it is distinguished using the least square method of recursion with forgetting factor Know power battery linear Ageing Model parameter to be identified when, need to identification start before preset power battery linear attenuation rail The number for the data point that the data length of mark, that is, linear attenuation track includes, and when the data length of preset linear attenuation track For L0When, that is to say, that it is preset when data length of the power battery in the decaying track in the linear attenuation stage of second stage is more than Value L0When, start the least square method of recursion with forgetting factor and the parameter to be identified of constructed linear Ageing Model is carried out Identification.Since each data point in the linear attenuation track of power battery corresponds to a charge and discharge cycles number, therefore, it can Think to be more than preset value L in the charge and discharge cycles number of linear decling phase when power battery0When, start with forgetting factor Least square method of recursion recognizes the parameter to be identified of constructed linear Ageing Model.When actually being recognized, L0 Value can be chosen according to actual conditions, in general, L0Bigger, the parameter for recognizing the linear Ageing Model obtained is more smart Really, but constructed the initial cycles number for the linear Ageing Model that the remaining life of power battery is estimated can also be got over Greatly, that is, the starting charge and discharge cycles number for estimating the remaining life of power battery more postpones.
Specific identification process is as follows:
First, Recursive Least Squares Estimation device of the initialization with forgetting factor,
Wherein,
θ indicates the parameter vector to be identified of the linear Ageing Model of power battery, and θ=[b0, b1]T, T representing matrixes turn It sets,
For the initial estimate of the parameter vector to be identified of the linear Ageing Model of power battery, setting
E expressions take average operation,
Q indicates the error co-variance matrix of the parameter vector estimated value to be identified of the linear Ageing Model of power battery,
Q0For the parameter vector estimated value to be identified of the linear Ageing Model of power battery error co-variance matrix just Initial value sets Q0=106I, and I is unit matrix;
Then, into the Recursive Least Squares Estimation device with forgetting factor input power battery in the linear decling phase The number i of linear attenuation data point, and i=1,2,3, L constantly acquires the new charge and discharge cycles data and more of power battery The estimated value of the parameter vector to be identified of the linear Ageing Model of new power batteryIt is new until collected power battery Charge and discharge cycles data are that (power battery is from linear attenuation rank for the corresponding data point number of charge and discharge cycles number at current time The charge and discharge cycles number carried out to current time is lighted in section starting) L, to which update obtains the linear Ageing Model of power battery Parameter vector to be identified estimated valueWherein, linear attenuation data point includes power battery in the linear attenuation data point When corresponding charge and discharge cycles number k and health status value SOHk, k=k0- 1+i, and charge and discharge cycles number k0For power battery Health status second stage the linear attenuation stage starting charge and discharge cycles number.At this point, description power battery is the The Ageing Model in the linear attenuation stage of two-stage is:
That is,
Specific renewal process is as follows:
As i=1, it is based on initialization as a result,
Update gain matrix K is obtained,
Wherein,
K1Indicate corresponding power battery the 1st data point of linear decling phase gain matrix,
Q0Indicate the parameter vector to be identified of the linear Ageing Model of power battery when corresponding 0th linear attenuation data point The error co-variance matrix of estimated value,
Indicate the input vector of the Recursive Least Squares Estimation device with forgetting factor, and
Indicate that the Recursive Least Squares Estimation device with forgetting factor is corresponding defeated in the 1st linear attenuation data point Incoming vector, andI.e.
The parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus batteryIt obtains
Wherein,
Indicate the parameter to be identified of the linear Ageing Model of power battery in the 1st data point of linear decling phase Vectorial estimated value,
Indicate the parameter to be identified of the linear Ageing Model of power battery in the 0th data point of linear decling phase Vectorial estimated value,
Y indicates the observation of the Recursive Least Squares Estimation device with forgetting factor, and y=SOHk,
y1Indicate observation of the Recursive Least Squares Estimation device with forgetting factor in the 1st linear attenuation data point Value,I.e.And the health status value of power batteryFor the healthy shape of power battery Initial value of the state in the linear attenuation stage of second stage;
The error co-variance matrix Q of the parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus battery, obtains It arrives
Wherein,
Q1Indicate the parameter vector to be identified of the linear Ageing Model of power battery when corresponding 1st linear attenuation data point The error co-variance matrix of estimated value;
μ indicates forgetting factor, and the value range of μ is (0,1).
As i=2, the update based on i=1 as a result,
Update gain matrix K is obtained,
Wherein,
K2Indicate corresponding power battery the 2nd data point of linear decling phase gain matrix,
Indicate that the Recursive Least Squares Estimation device with forgetting factor is corresponding defeated in the 1st linear attenuation data point Incoming vector, andCharge and discharge cycles number k0+ 1 is line of the health status in second stage of power battery Second charge and discharge cycles number of property decling phase;
The parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus batteryIt obtains
Wherein,
Indicate the parameter to be identified of the linear Ageing Model of power battery in the 2nd data point of linear decling phase Vectorial estimated value,
y2Indicate observation of the Recursive Least Squares Estimation device with forgetting factor in the 2nd linear attenuation data point Value,And the health status value of power batteryFor power battery health status in second-order The second value in the linear attenuation stage of section;
The error co-variance matrix Q of the parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus batteryi, obtain It arrives
Wherein,
Q2Indicate the parameter vector to be identified of the linear Ageing Model of power battery when corresponding 2nd linear attenuation data point The error co-variance matrix of estimated value,
μ indicates forgetting factor, and the value range of μ is (0,1).
And so on,
As i=i, the update result based on i=i-1
Gain matrix K is updated, is obtained
Wherein,
KiIndicate corresponding power battery i-th of data point of linear decling phase gain matrix,
Qi-1Indicate the parameter to be identified of the linear Ageing Model of power battery when corresponding (i-1)-th linear attenuation data point The error co-variance matrix of vectorial estimated value,
Indicate that the Recursive Least Squares Estimation device with forgetting factor is corresponding defeated in i-th of linear attenuation data point Incoming vector, and
The parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus batteryAnd
Wherein,
Indicate the parameter to be identified of the linear Ageing Model of power battery in i-th of data point of linear decling phase Vectorial estimated value,
Indicate (i-1)-th data of the parameter to be identified in the linear decling phase of the linear Ageing Model of power battery Vectorial estimated value when point,
yiIndicate observation of the Recursive Least Squares Estimation device with forgetting factor in i-th of linear attenuation data point Value, and
The error co-variance matrix Q of the parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus battery, and
Wherein,
QiIndicate the parameter vector to be identified of the linear Ageing Model of power battery when corresponding i-th of linear attenuation data point The error co-variance matrix of estimated value,
μ indicates forgetting factor, and the value range of μ is (0,1).
......
As i=L, the update based on i=L-1 as a result,
Gain matrix K is updated, is obtained
Wherein,
KLIndicate corresponding power battery the l-th data point of linear decling phase gain matrix,
QL-1Indicate the parameter to be identified of the linear Ageing Model of power battery when corresponding the L-1 linear attenuation data point The error co-variance matrix of vectorial estimated value,
Indicate that the Recursive Least Squares Estimation device with forgetting factor is corresponding defeated in l-th linear attenuation data point Incoming vector, and
The parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus batteryAnd
Wherein,
Indicate the parameter to be identified of the linear Ageing Model of power battery in the l-th data point of linear decling phase Vectorial estimated value,
Indicate the L-1 data of the parameter to be identified in the linear decling phase of the linear Ageing Model of power battery Vectorial estimated value when point,
yLIndicate observation of the Recursive Least Squares Estimation device with forgetting factor in l-th linear attenuation data point Value, and
The error co-variance matrix Q of the parameter vector estimated value to be identified of the linear Ageing Model of renewal impetus battery, and
Wherein,
QLIndicate the parameter vector to be identified of the linear Ageing Model of power battery when corresponding l-th linear attenuation data point The error co-variance matrix of estimated value,
μ indicates forgetting factor, and the value range of μ is (0,1).
At no point in the update process, the value range of μ is preferably (0.95,1), in this way, it is ensured that carries the recursion of forgetting factor The robust property of LS estimator.
Finally, according toPick out the parameter to be identified of the linear Ageing Model of power battery.
In the art, as the real time health state SOH of power batterykWhen less than 80%, that is, think the power battery Service life ends.Therefore when the remaining life of power battery, that is, remaining charge and discharge cycles number is p, then the power battery is in the longevity The health status value of life cut-off time is represented byAnd
And then estimate to obtain the remaining charge and discharge cycles number i.e. remaining life of the power battery.
The method of the remaining life of the power battery of this estimation electric vehicle is by way of recursion to power battery Online data is iterated operation, accurately estimates out the remaining life of power battery, effectively promotes the stability of estimation results And reliability.In addition, this method of estimation, is both not required to a large amount of off-line datas of acquisition power battery, is also not required to online data Repetition training is carried out, it is simple and convenient, greatly reduce the cost for the remaining life for estimating power battery.
In the following, for including four kinds of power batteries that 0.5C, 1C, 1.5C, 2C recycle discharge-rate successively, to the present invention The method of the remaining life of the power battery of the estimation electric vehicle of proposition is verified.
First, to including that four power batteries of 0.5C, 1C, 1.5C, 2C cycle discharge-rate carry out accelerated aging successively Experiment obtains the decaying track of the health status value SOH of four power batteries successively such as Fig. 1 (a), 1 (b), 1 (c) and 1 (d) It is shown, and solid line in the 4 width figure indicates the practical power battery obtained that measures in corresponding charge and discharge cycles number Health status value SOH, the actual measured value of the health status of corresponding power battery;Dotted line then indicates the power that linear fit obtains Battery corresponds to the linear fit value of the health status of power battery in the decaying track in the linear attenuation stage of second stage, and When carrying out linear fit, the linear fit characteristic such as table 1 of the health status value SOH of power battery in the decaying track of second stage It is shown.
Linear fit characteristics of the health status value SOH of 1 power battery of table in the decaying track of second stage
Multiplying power R squares Root-mean-square value (%)
0.5C 0.9975 0.19
1C 0.9972 0.19
1.5C 0.9852 0.46
2C 0.9946 0.31
Wherein,
The R squares of health status value SOH for representing power battery is in the linearity of the decaying track of second stage, and R squares 0 And change between 1, and R squares closer to 1, shows that the linearity of data is better;
Root-mean-square value represents the gap between linear fit value and actual measured value, and root-mean-square value is smaller, illustrates linear Match value is closer with actual measured value, and real data linear attenuation feature is more apparent.
As shown in Figure 1, when 0.5C, 1C, 1.5C, 2C are recycled under discharge-rate, power battery is in the linear decling phase The health status value SOH of starting point0Predeterminable is 92%, at this point, the linear fit value of the health status value SOH of power battery It is very close with actual measured value, especially when it is 0.5C, 1C and 2C to recycle discharge-rate, the health status value of power battery The linear fit value of SOH and actual measured value are about the same;When it is 1.5C to recycle discharge-rate, the health status of power battery Though the situation that the registration between the linear fit value and actual measured value of value SOH is not so good as under other cycle discharge-rates is good, It is still to show preferable registration therebetween.In addition, as shown in Table 1, when cycle discharge-rate is 0.5C, 1C and 2C When, R squares all 0.99 or more, root mean square is all below 0.5%;When it is 1.5C to recycle discharge-rate, R squares close 0.99, root mean square is also below 0.5%.It can be seen that when setting 92% is for the second stage of the decaying track of power battery When the health status value of the starting point in linear attenuation stage, the linear attenuation stage of the health status of power battery is with very good Linear attenuation characteristic, and the corresponding charge and discharge cycles number of starting point in the linear attenuation stage of four power batteries is followed successively by 100,123,142 and 172.
Due to L0Value can be chosen according to actual conditions, the remaining life of power battery is carried out in order to earlier pre- Estimate, and obtain accurate remaining life estimated value, inventor is when verifying by L0It is set as 50.Due to forgetting factor μ's Value range is (0,1), and forgetting factor μ values are bigger, and least square method of recursion estimator gets over the dependence of historical information By force, robustness is better;Forgetting factor μ values are smaller, and least square method of recursion estimator is stronger to the dependence of new information, to letter The tracing property of breath is better.Inventor integrates the robustness and tracing property of least square method of recursion, and the value range of μ is set as (0.95,1).When being estimated, any one value in the value range can be chosen according to practical application.In confirmatory experiment In, inventor chooses value of the mediant as forgetting factor μ between 0.95 and 1, i.e. μ=0.97.
In the health status value SOH of the starting point in the linear attenuation stage for the health status for determining power battery0It is 92%, The data length L of preset linear attenuation track0It is 50, after forgetting factor is 0.97, most using the recursion with forgetting factor Small square law recognizes the parameter to be identified of the linear Ageing Model of power battery, and obtains power electric according to identification result The linear Ageing Model in pond, so using the linear Ageing Model estimate out cycle discharge-rate be followed successively by 0.5C, 1C, 1.5C, The remaining life of four power batteries of 2C, as shown in Fig. 2, and estimation results shown in Fig. 2 are every 10 charge and discharge cycles Corresponding remaining life estimated value, wherein error bounds differ 50 charge and discharges with the actual value of the remaining life of power battery Electricity cycle, the discreet value of the remaining life of power battery are fallen into error boundary, you can think the remaining life of the power battery Discreet value be accurate predicted value.As shown in Figure 2, the power battery for being 0.5C and 1C for cycle discharge-rate, the present invention The method of estimation of proposition can be realized dynamic from the predicting residual useful life value for estimating starting point and just obtaining accurate power battery The prediction accurate earlier of the remaining life of power battery, and with collected real-time remaining capacity value CmNumber increase, Estimate that the estimated value of the remaining life of obtained power battery is restrained to the actual value of the remaining life of the power battery rapidly, and It is almost the same with the actual value of the remaining life of power battery in the remaining life end-of-cycle of power battery.It discharges for cycle Multiplying power is the power battery of 1.5C and 2C, and method of estimation proposed by the present invention remains able to realize carrying for power battery remaining life Preceding prediction, wherein the power battery for being 1.5C for cycle discharge-rate, the estimated value of remaining life are filled close to 350 It is fallen into error boundary when discharge cycles, and with collected real-time remaining capacity value CmNumber increase, estimation obtains The estimated value of remaining life of power battery converge on actual value rapidly, and the remaining life end-of-cycle of power battery with The actual value of the remaining life of power battery is almost the same;The power battery for being 2C for cycle discharge-rate, remaining life Estimated value fallen into error boundary when close to 250 charge and discharge cycles, and with collected real-time remaining capacity value Cm Number increase, the estimated value of the remaining life for the power battery estimated converges on rapidly actual value, and in power electric The remaining life end-of-cycle in pond is almost the same with the actual value of the remaining life of power battery.To sum up, cycle is put Electric multiplying power is the power battery of 0.5C, 1C, 1.5C and 2C, and method of estimation of the present invention can calculate to a nicety out power battery Remaining life.
In addition, the present invention also proposes a kind of power battery management system, the power battery management system is using above-mentioned arbitrary A kind of method of remaining life of the power battery of estimation electric vehicle estimates the remaining life row of the power battery of electric vehicle Meter.Such power battery management system is relatively existing when the remaining life of the power battery to electric vehicle is estimated Mainstream power battery management system, estimated accuracy is high, safe and reliable and at low cost.

Claims (6)

1. a kind of method of the remaining life of the power battery of estimation electric vehicle, which is characterized in that this method includes following step Suddenly:
Step S1, in the power battery course of work, the charging and discharging currents and terminal voltage of the power battery are acquired in real time, And On-line Estimation goes out the real-time remaining capacity value C of the power batterym
Step S2, according to the real-time remaining capacity value C of the power batterymWith the initial capacity value C of the power battery0, calculate Go out the real time health state SOH of the power battery, and
Step S3, to the inflection point pair of the decaying track of the real time health state value SOH of the power battery and the power battery The health status value SOH answered0It is compared, when the real time health state value SOH of the power battery is less than the power battery The corresponding health status SOH of decaying track inflection point0When, the real time health state value SOH of the power battery0Corresponding charge and discharge Electric cycle-index k0For the starting point in the linear attenuation stage of the power battery;
Step S4, the linear aging of the power battery is established according to the decaying track in the linear attenuation stage of the power battery Model:SOHk=b0+b1·k+εk, εk~N (0, σ2),
Wherein,
K indicates the power battery in current time corresponding charge and discharge cycles number,
b0, b1Indicate the parameter to be identified of the linear Ageing Model of the power battery,
εkIt is 0 for mean value, variance σ2Independent normal stochastic error,
N indicates normal distribution,
And according to the charging and discharging currents and terminal voltage value of the power battery acquired in real time, on-line identification obtains described linear old Change the parameter to be identified of model;
Step S5, estimate the remaining charge and discharge cycles number of the power battery, and
Wherein,
The power battery is indicated in cut-off time in service life corresponding health status value,
L indicates that the charge and discharge cycles number carried out to current time is lighted in starting to the power battery from the linear attenuation stage,
P indicates the power battery from current time to the remaining charge and discharge cycles number of cut-off time in service life.
2. the method for the remaining life of the power battery of estimation electric vehicle according to claim 1, which is characterized in that In the step S3, pair accelerated life test is carried out with the power battery of the power battery same model, obtains the power The starting point in the linear attenuation stage of battery.
3. the method for the remaining life of the power battery of estimation electric vehicle according to claim 1 or 2, feature exist In in the step S4, waiting for for the linear Ageing Model is picked out using the least square method of recursion with forgetting factor Identified parameters.
4. the method for the remaining life of the power battery of estimation electric vehicle according to claim 3, which is characterized in that distinguish Knowledge process is as follows:
First, Recursive Least Squares Estimation device of the initialization with forgetting factor,
Wherein,
θ indicates the parameter vector to be identified of the linear Ageing Model of the power battery, and θ=[b0, b1]T, T representing matrixes turn It sets,
For the initial estimate of the parameter vector to be identified of the linear Ageing Model of the power battery, setting
E expressions take average operation,
Q indicates the error co-variance matrix of the parameter vector estimated value to be identified of the linear Ageing Model of the power battery,
Q0For the parameter vector estimated value to be identified of the linear Ageing Model of the power battery error co-variance matrix it is initial Value sets Q0=106I, and I is unit matrix;
Then, the power battery is inputted into the Recursive Least Squares Estimation device with forgetting factor in linear attenuation rank The number i of the linear attenuation data point of section, and i=1,2,3, L, the linear attenuation data point includes the power battery Charge and discharge cycles number k and the power battery corresponding health status value SOH when charge and discharge cycles number is kk, and k= k0- 1+i,
Update gain matrix Ki, and
Update the parameter vector estimated value to be identified of the linear Ageing Model of the power batteryAnd
Update the error co-variance matrix Q of the parameter vector estimated value to be identified of the linear Ageing Model of the power batteryi, and
Wherein,
Indicate the input vector of the Recursive Least Squares Estimation device with forgetting factor, and
Y indicates the observation of the Recursive Least Squares Estimation device with forgetting factor, and y=SOHk,
μ indicates forgetting factor, and the value range of μ is (0,1);
Finally, according toPick out the parameter to be identified of the linear Ageing Model of the power battery.
5. the method for the remaining life of the power battery of estimation electric vehicle according to claim 4, which is characterized in that institute The value range for stating forgetting factor μ is (0.95,1).
6. a kind of method using described in any one of claim 1-5 to the remaining life of the power battery of electric vehicle into The power battery management system of row estimation.
CN201810349151.6A 2018-04-18 2018-04-18 Method for estimating residual life of power battery of electric automobile Active CN108490365B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810349151.6A CN108490365B (en) 2018-04-18 2018-04-18 Method for estimating residual life of power battery of electric automobile

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810349151.6A CN108490365B (en) 2018-04-18 2018-04-18 Method for estimating residual life of power battery of electric automobile

Publications (2)

Publication Number Publication Date
CN108490365A true CN108490365A (en) 2018-09-04
CN108490365B CN108490365B (en) 2020-07-03

Family

ID=63313728

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810349151.6A Active CN108490365B (en) 2018-04-18 2018-04-18 Method for estimating residual life of power battery of electric automobile

Country Status (1)

Country Link
CN (1) CN108490365B (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109991546A (en) * 2019-03-29 2019-07-09 深圳猛犸电动科技有限公司 A kind of battery parameter acquisition methods, device and terminal device
CN112231830A (en) * 2020-09-30 2021-01-15 浙江大学 Hybrid power vehicle multi-objective optimization control method based on adaptive equivalent factor
WO2021056864A1 (en) * 2019-09-26 2021-04-01 广东小鹏汽车科技有限公司 Method and apparatus for estimating residual energy of battery, and vehicle
CN113433466A (en) * 2021-05-28 2021-09-24 北京元代码科技有限公司 New energy automobile power battery detection method and system
CN113447824A (en) * 2021-06-28 2021-09-28 三一重型装备有限公司 Method and device for estimating maximum charging and discharging current of battery and storage medium
CN113928330A (en) * 2021-10-21 2022-01-14 重庆大学 Pure electric vehicle speed trajectory optimization control method
CN113987909A (en) * 2021-09-18 2022-01-28 广东电网有限责任公司广州供电局 Oil paper insulation aging prediction method and device, computer equipment and storage medium
CN114460484A (en) * 2021-07-15 2022-05-10 崔跃芹 Rechargeable battery service life prediction method and device based on accumulated loss
CN117554846A (en) * 2024-01-12 2024-02-13 烟台海博电气设备有限公司 Lithium battery life prediction method and system considering constraint conditions

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102778653A (en) * 2012-06-20 2012-11-14 哈尔滨工业大学 Data-driven lithium ion battery cycle life prediction method based on AR (Autoregressive) model and RPF (Regularized Particle Filtering) algorithm
US8427166B2 (en) * 2006-03-28 2013-04-23 Gem Power, Llc State of health recognition of secondary batteries
CN103778280A (en) * 2014-01-08 2014-05-07 上海交通大学 Method for remain useful life prognostic of lithium ion battery with model active updating strategy
CN107069122A (en) * 2017-04-01 2017-08-18 山东省科学院自动化研究所 A kind of Forecasting Methodology of electrokinetic cell remaining life

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8427166B2 (en) * 2006-03-28 2013-04-23 Gem Power, Llc State of health recognition of secondary batteries
CN102778653A (en) * 2012-06-20 2012-11-14 哈尔滨工业大学 Data-driven lithium ion battery cycle life prediction method based on AR (Autoregressive) model and RPF (Regularized Particle Filtering) algorithm
CN103778280A (en) * 2014-01-08 2014-05-07 上海交通大学 Method for remain useful life prognostic of lithium ion battery with model active updating strategy
CN107069122A (en) * 2017-04-01 2017-08-18 山东省科学院自动化研究所 A kind of Forecasting Methodology of electrokinetic cell remaining life

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109991546A (en) * 2019-03-29 2019-07-09 深圳猛犸电动科技有限公司 A kind of battery parameter acquisition methods, device and terminal device
CN109991546B (en) * 2019-03-29 2021-08-13 深圳猛犸电动科技有限公司 Battery parameter acquisition method and device and terminal equipment
WO2021056864A1 (en) * 2019-09-26 2021-04-01 广东小鹏汽车科技有限公司 Method and apparatus for estimating residual energy of battery, and vehicle
CN112231830A (en) * 2020-09-30 2021-01-15 浙江大学 Hybrid power vehicle multi-objective optimization control method based on adaptive equivalent factor
CN113433466A (en) * 2021-05-28 2021-09-24 北京元代码科技有限公司 New energy automobile power battery detection method and system
CN113447824A (en) * 2021-06-28 2021-09-28 三一重型装备有限公司 Method and device for estimating maximum charging and discharging current of battery and storage medium
CN114460484A (en) * 2021-07-15 2022-05-10 崔跃芹 Rechargeable battery service life prediction method and device based on accumulated loss
CN114460484B (en) * 2021-07-15 2024-01-09 崔跃芹 Rechargeable battery life prediction method and device based on accumulated wear quantity
CN113987909A (en) * 2021-09-18 2022-01-28 广东电网有限责任公司广州供电局 Oil paper insulation aging prediction method and device, computer equipment and storage medium
CN113928330A (en) * 2021-10-21 2022-01-14 重庆大学 Pure electric vehicle speed trajectory optimization control method
CN117554846A (en) * 2024-01-12 2024-02-13 烟台海博电气设备有限公司 Lithium battery life prediction method and system considering constraint conditions
CN117554846B (en) * 2024-01-12 2024-03-22 烟台海博电气设备有限公司 Lithium battery life prediction method and system considering constraint conditions

Also Published As

Publication number Publication date
CN108490365B (en) 2020-07-03

Similar Documents

Publication Publication Date Title
CN108490365A (en) A method of the remaining life of the power battery of estimation electric vehicle
CN106054084B (en) A kind of power battery SOC estimation method
CN111581904B (en) Lithium battery SOC and SOH collaborative estimation method considering cycle number influence
CN104502858B (en) Electrokinetic cell SOC methods of estimation and system based on backward difference discrete model
CN109061520A (en) A kind of power battery health and power rating estimation on line method and system
CN105510829B (en) A kind of Novel lithium ion power battery SOC methods of estimation
CN109061506A (en) Lithium-ion-power cell SOC estimation method based on Neural Network Optimization EKF
CN110261778B (en) SOC estimation algorithm of lithium ion battery
CN108445402A (en) A kind of lithium-ion-power cell state-of-charge method of estimation and system
CN106353687A (en) Assessment method of lithium battery health status
CN105974327A (en) Lithium battery pack SOC prediction method based on neural network and UKF
CN106443474A (en) Method for rapidly recognizing life decline characteristic of power battery system
CN105116344B (en) Based on binary-coded battery open circuit voltage evaluation method
CN103529398A (en) Online lithium ion battery SOC (state of charge) estimation method based on extended Kalman filter
CN106154163B (en) Battery life state identification method
CN105929338B (en) A kind of method and its application measuring battery status
CN105572596B (en) Lithium battery SOC estimation method and system
CN110554321B (en) Method for detecting SOC (state of charge) of retired power battery in real time
CN109901072B (en) Retired battery parameter detection method based on historical data and laboratory test data
CN108445422B (en) Battery state of charge estimation method based on polarization voltage recovery characteristics
CN109917299B (en) Three-layer filtering estimation method for state of charge of lithium battery
CN109358293A (en) Lithium ion battery SOC estimation method based on IPF
CN108872861A (en) A kind of method of online evaluation cell health state
CN112083331A (en) Fusion method for improving lithium ion battery state of charge estimation precision
CN115980588A (en) Lithium ion battery health state estimation method based on self-encoder extraction features

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
GR01 Patent grant
GR01 Patent grant