CN106055775B - A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model - Google Patents
A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model Download PDFInfo
- Publication number
- CN106055775B CN106055775B CN201610363499.1A CN201610363499A CN106055775B CN 106055775 B CN106055775 B CN 106055775B CN 201610363499 A CN201610363499 A CN 201610363499A CN 106055775 B CN106055775 B CN 106055775B
- Authority
- CN
- China
- Prior art keywords
- secondary cell
- particle
- formula
- charge
- battery
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/36—Circuit design at the analogue level
- G06F30/367—Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods
Landscapes
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model, the present invention is to solve service life of secondary cell of the tradition based on particle filter and predict based entirely on data-driven, the defect for ignoring prediction object Characteristics, leads to the problem to the prediction result accuracy difference in electrochemical power source service life.It is new state equation that training stage, which obtains state variable with the regression equation that charge and discharge cycles number changes with the true value that particle filter method tracks inside battery state variable,；State variable estimate when forecast period calculates unknown charge and discharge cycles using new state equation, generate multiple particles, it substitutes into observational equation and obtains the estimated value of multiple capacity observed quantities, with the median of multiple capacity observation estimated values oppose certain following charge and discharge cycles when battery capacity prediction, when reaching preset battery capacity lower limit, the difference of cycle-index used in cycle-index corresponding to the capacity predicted value and training stage is the available cycles left number of battery.
Description
Technical field
The present invention relates to a kind of by secondary cell (including lithium ion battery, lead-acid battery, hereinafter referred to as battery) mechanism mould
The battery life predicting new method that type emulation technology is combined with particle filter algorithm.Belong to equipment dependability field.
Background technique
In recent years, the secondary rechargeable batteries such as lead-acid battery and lithium ion battery obtain in fields such as electric car, smart grids
Obtained extensive use.From the angle used, the life problems of battery have become the bottle for restricting electric car, smart grid development
Neck problem.
The service life of Accurate Prediction battery is the basic demand of the battery system maintenance based on state, for improving battery system
Reliability, the save the cost of system are most important.The life-span prediction method of battery can be divided into three classes: " being based on agine mechaism " needs
It is to be understood that causing such as catalyst effective area of cell degradation to reduce, can be reduced with conductive ion concentration, electrode passivation film increases
Aging mechanism such as long, and modeling to it, the modeling of single aging mechanism is with regard to extremely complex, phase again between various aging modes
Mutual coupling, so the life-span prediction method based on agine mechaism is difficult to realize；" being based on data-driven ", according to going through for battery capacity
History data variation trend predicts battery performance in conjunction with nonlinear regression, Kalman filtering, particle filter scheduling algorithm, this
Kind method ignores the physical significance and battery object of data, is difficult the precision of prediction obtained；" being based on feature ", in conjunction with reflection
Battery life surveys feature prediction battery life, and usual this feature is relatively difficult to choose, and characteristic quantity and battery capacity
Between connection be difficult to quantify.
The thought of particle filter is based on monte carlo method, it is to indicate probability using particle collection, can be used in any
On the state-space model of form.Its core concept is stochastic regime particle by extracting from posterior probability come expression status
Variable distribution, is a kind of sequence importance sampling method.In simple terms, particle filter method refers to by finding one group in state sky
Between the random sample propagated probability density function is carried out approximate, integral operation is replaced with sample average, to obtain state most
The process of small variance distribution.Here sample refers to particle, and any form can be approached when sample size levels off to infinity
Probability density distribution.Particle filter has the characteristics that imparametrization, and random quantity must when getting rid of solution Nonlinear Filtering Problem
The restriction of Gaussian Profile must be met, distribution more wider than Gauss model can be expressed, also had to the nonlinear characteristic of variable parameter
Stronger modeling ability.Therefore, particle filter can relatively accurately express the posterior probability based on observed quantity and control amount point
Cloth can obtain more accurate system state estimation result.
Summary of the invention
The present invention is to solve service life of secondary cell of the tradition based on particle filter and predict based entirely on data-driven, suddenly
Defect depending on predicting object Characteristics, leads to the problem to the prediction result accuracy difference of battery life.A kind of grain is now provided
The service life of secondary cell prediction technique that son filtering is combined with mechanism model.
A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model, it includes the following contents:
Step 1: the mechanism model of building secondary cell, the mechanism model of the secondary cell can simulate any electric current
The curve that the charging/discharging voltage of battery changes over time when condition；
Step 2: the training stage: the secondary cell in step 1 is subjected to aging for a period of time under normal use operating condition,
The charging and discharging curve in dynamic operation condition off-line measurement secondary cell ageing process is utilized at interval of fixed charge and discharge cycles number,
The voltage obtained at this time is practical secondary cell output voltage U,
Same dynamic operation condition electric current is inputted to the emulation of the mechanism model of secondary cell, is exported with model emulationIt goes to simulate
Reality output U of the secondary cell in the different ageing steps, using genetic algorithm or least square method, according to objective function realization pair
The identification of secondary cell model parameter collection P, the multiple parameters collection P for secondary cell each ageing step that identification is obtained is as instruction
Practice data,
Selected from the multiple parameters collection P that uses of training L mechanism model parameter relevant to ageing process as state to
Measuring X, wherein L is positive integer, using the battery capacity Q under actual load current conditions as observed quantity, same load current conditions
Under battery mechanism model emulation and conversion capacity estimation valueProcess as observational equation, made always using particle filter algorithm
The estimated value of the state vector in each stage during changeClose to true value X；
Step 3: prediction process: being estimated using the state vector passed through in particle filter algorithm training process in step 2
Value sequence obtains regression equation of the state vector X about cycle-index k, in this, as new shape with the method for polynomial regression
State equation obtains the estimated value of state vector by new state equation when k is some following charge and discharge cycles numberSubstitute into equation:
In formula, k is cycle-index, XP,i,j(k) i-th of component for indicating j-th of particle, obeying mean value isVariance
For σw,iNormal distribution, 1≤i≤L, 1≤j≤M, wi(k) the systematic procedure noise for being state variable Xi,
The multiple particles for meeting Gaussian Profile are obtained, multiple particles are substituted into the observational equation in step 2, to obtain
The medians of multiple observed quantity estimated values preset when prediction capacity reaches as the predicted value to the following battery capacity
Battery capacity lower limit when, the difference of cycle-index used in the training stage is that battery can in corresponding cycle-index and step 2
Cycles left number, to realize the prediction to secondary cell remaining life.
The invention has the benefit that utilizing particle filter method tracking inside battery state vector in the training stage
True value, and the regression equation changed with obtained state vector with charge and discharge cycles number is new state equation.It is predicting
Stage, the estimated value of state variable when being calculated unknown charge and discharge cycles using new state equation are generated more on this basis
A particle substitutes into the estimated value that multiple capacity observed quantities are obtained in observational equation, in multiple capacity observation estimated values respectively
The prediction of battery capacity when digit is as to certain following charge and discharge cycles, when the predicted value of battery capacity reaches preset
When battery capacity lower limit, the difference of cycle-index used in cycle-index corresponding to the capacity predicted value and training stage is electricity
The available cycles left number in pond.It is for predicting the service life of electrochemical power source.
Mechanism electrochemical model is combined with particle filter algorithm for the first time, applied to the life prediction of secondary cell, is adopted
The life prediction result phase of secondary cell is obtained with using existing method with the prediction result that this method obtains service life of secondary cell
It is reduced within 10% than prediction error.This method breaches traditional particle filter life prediction based entirely on data-driven
Method.This method is joined using mechanism model simulated program as observer, with the mechanism model regularly changing with cell degradation
Number is used as state variable, improves to conventional particle filter forecasting method.Compared to conventional particle filtering method, this method has
The characteristics of observational equation precision height, state variable explicit physical meaning, it can be realized the Accurate Prediction to remaining battery life.It can
The life prediction of secondary cell for different principle.
Detailed description of the invention
Fig. 1 is that a kind of service life of secondary cell that particle filter is combined with mechanism model described in specific embodiment one is pre-
The flow chart of survey method；
Fig. 2 is certain lead-acid battery DST operating condition current curve diagram；
Fig. 3 is certain lead-acid battery DST operating condition voltage curve.
Specific embodiment
Specific embodiment 1: illustrating present embodiment, a kind of grain described in present embodiment referring to figs. 1 to Fig. 3
The service life of secondary cell prediction technique that son filtering is combined with mechanism model, it includes the following contents:
Step 1: the mechanism model of building secondary cell, the mechanism model of the secondary cell can simulate any electric current
The curve that the charging/discharging voltage of battery changes over time when condition；
Step 2: the training stage: the secondary cell in step 1 is subjected to aging for a period of time under normal use operating condition,
The charging and discharging curve in dynamic operation condition off-line measurement secondary cell ageing process is utilized at interval of fixed charge and discharge cycles number,
The voltage obtained at this time is practical secondary cell output voltage U,
Same dynamic operation condition electric current is inputted to the emulation of the mechanism model of secondary cell, is exported with model emulationIt goes to simulate
Reality output U of the secondary cell in the different ageing steps, using genetic algorithm or least square method, according to objective function realization pair
The identification of secondary cell model parameter collection P, the multiple parameters collection P for secondary cell each ageing step that identification is obtained is as instruction
Practice data,
Selected from the multiple parameters collection P that uses of training L mechanism model parameter relevant to ageing process as state to
Measuring X, wherein L is positive integer, using the battery capacity Q under actual load current conditions as observed quantity, same load current conditions
Under battery mechanism model emulation and conversion capacity estimation valueProcess as observational equation, made always using particle filter algorithm
The estimated value of the state vector in each stage during changeClose to true value X；
Step 3: prediction process: being estimated using the state vector passed through in particle filter algorithm training process in step 2
Value sequence obtains regression equation of the state vector X about cycle-index k, in this, as new shape with the method for polynomial regression
State equation obtains the estimated value of state vector by new state equation when k is some following charge and discharge cycles numberSubstitute into equation:
In formula, k is cycle-index, XP,i,j(k) i-th of component for indicating j-th of particle, obeying mean value isVariance
For σw,iNormal distribution, 1≤i≤L, 1≤j≤M, wi(k) the systematic procedure noise for being state variable Xi,
The multiple particles for meeting Gaussian Profile are obtained, multiple particles are substituted into the observational equation in step 2, to obtain
The medians of multiple observed quantity estimated values preset when prediction capacity reaches as the predicted value to the following battery capacity
Battery capacity lower limit when, the difference of cycle-index used in the training stage is that battery can in corresponding cycle-index and step 2
Cycles left number, to realize the prediction to secondary cell remaining life.
In present embodiment, for secondary cell, the time of aging is such as 20 circulations.Secondary cell is in normal use
Aging is carried out for a period of time at interval of fixed charge and discharge cycles number under operating condition, according to objective function, measures multiple charge and discharge
Multiple groups parameter set under circulation, selects the smallest several groups of parameter sets as training data, makes the training using particle filter algorithm
State vector estimated value in dataClose to true value X after undergoing multiple cycle charge-discharge；It is instructed using particle filter algorithm
State vector during white silk estimates value sequence, with the method for polynomial regression, obtains L new mechanism model parameter through excessive
State equation X (k) after secondary circulation.
One, electrochemical mechanism models
Electrochemical mechanism model herein refers to its performance simulation model, i.e. mode input is that the charge or discharge of battery are electric
Stream, model output are the curve that corresponding end voltage changes over time.
Electrochemical model includes to battery electrode thermodynamics reversible voltage (open-circuit voltage), liquid phase diffusion and migration, solid phase
The mathematical description of the processes such as diffusion, electrochemical reaction dynamics is usually expressed as partial differential equation and its boundary condition, initial value item
The form of part can be iterated solution by finite difference method.The input of model, output relation can indicate are as follows:
U (t)=f [I (t), P (k)] (1)
Wherein, Function Mapping f () is that the number of end voltage is calculated by mechanism model to constant current I (t) and parameter set P (k)
It is worth simulation process, describes under specific operation voltage U with the variation of charge and discharge time；K is charge and discharge cycles number, it is believed that parameter
Collection P changes with the increase of charge and discharge number.
Because the input of performance simulation model is present battery status and applying working condition, export as external voltage-measurable, because
This can be used as the observational equation of battery system.In problem of aging, the observed quantity of cell degradation is usually capacity, can be in electricity
Pond passes through discharge voltage profile of formula (1) artificial battery under the conditions of actual load under conditions of completely filling, according to discharge voltage
Cut off determine electric discharge cut-off time, since electric discharge to electric discharge end, integrated current over time (current integration method) is obtained
Obtain the capacity of battery.The calculating process of capacity can be described as
Q (k)=q [I (t), P (k)] (2)
Wherein, Q (k) is the estimated value of capacity, and I (t) is electric current used by measurement capacity, and P (k) is model parameter collection, q
[] indicates the process that discharge capacity is calculated according to discharge curve and discharge time.
Two, prepare training data
In order to obtain battery in the parameter set of different ageing steps, need to test battery in the different ageing steps of battery
Charging and discharging curve.The operating condition of test charging and discharging curve can choose ambulatory stress test operating condition DST (DynamicStress
Test), the operating condition contain reflection battery various processes the case where, the voltage of acquisition, electric current, electricity data collection abundant information,
The robustness of parameter identification result is stronger.Typical DST operating condition current curve and corresponding certain lead-acid battery voltage curve such as Fig. 2
With shown in Fig. 3.
The target of parameter identification is one group of parameter set of selection, when inputting same electric current, so that mechanism model emulation electricity
Pressure outputError between actual cell voltage output U is minimum, shown in objective function such as formula (3).
Wherein, I (t) is load current；P is parameter set to be identified；S is the search space of parameter set；N is the voltage that is taken
Change over time the data points on curve.
Parameter identification can be realized with genetic algorithm or least square method.
Three, some concepts of particle filter algorithm
(1) state variable
According to the parameter set of training stage and its variation, select with L closely related mechanism model parameter of aging as
State variable is denoted as state vector X.Other parameters fixation takes the average value repeatedly recognized.
Wherein, X1~XLFor ageing-related in model parameter collection P L parameter.
(2) observed quantity
The observed quantity used is the capacity of battery, measured value Q, and estimated value is
(3) observational equation
Battery terminal voltage simulation process formula (1) and calculation of capacity process formula (2) are combined, specific charge and discharge are obtained
The estimated value of observed quantity under operating condition I (t)This observation process is indicated in observational equation formula (5) with h [].In addition,
The calculating of observation is also contemplated that plus observation noise, it may be assumed that
Wherein, k is cycle-index, and v is observation noise, obey mean value be 0, variance σvGaussian Profile.
(4) state equation
In the training stage, how is the rule of unclear state variable variation, and state equation can be regarded as to following recursion
Relational expression:
K is cycle-index in formula；It is kth time circulation to state variable XiEstimated value；wiIt (k) is state variable Xi
Systematic procedure noise, obey mean value be 0, variance σw,iGaussian Profile.
Specific embodiment 2: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, in present embodiment, in step 1, and the machine of secondary cell
Manage model are as follows:
U (t)=f [I (t), P (k)] (formula 2),
In formula, I (t) is to constant current, and f is Function Mapping, and P (k) is the parameter set of secondary cell, and k is charge and discharge cycles
Number, parameter set P change with the increase of charge and discharge number k, and U (t) is the external voltage-measurable of secondary cell.
Specific embodiment 3: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, in present embodiment, in step 2, and objective function are as follows:
In formula, I (t) is to constant current, and P is parameter set to be identified, and S is the search space of parameter set, and N is the voltage that is taken
The data points on curve are changed over time,For mechanism model simulation data voltage, U is actual battery output voltage.
Specific embodiment 4: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, in present embodiment, in step 2, and state vector X are as follows:
In formula, X1~XLFor ageing-related in model parameter collection P L parameter.
Specific embodiment 5: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, in present embodiment, in step 2, battery capacity Q's
Equation are as follows:
Q (k)=q [I (t), P (k)] (formula 5),
In formula, I (t) is electric current used by measurement capacity, and P (k) is model parameter collection, and q [] indicates bent according to electric discharge
Line and discharge time calculate the process of discharge capacity；
In step 2, observational equation are as follows:
In formula,For the estimated value of the observed quantity under specific charge and discharge operating condition I (t), k is charge and discharge cycles number, and v is to see
Survey noise, obey mean value be 0, variance σvGaussian Profile.
Specific embodiment 6: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, and in present embodiment, in step 2, utilizes particle filter
Algorithm makes the estimated value of the state vector in each stage in ageing processClose to the detailed process of true value X are as follows:
Step A1, particle initialization is carried out using particle filter algorithm: setting state vector XiProcess-noise variance
σw,i, 1≤i≤L；Observation noise variances sigma is setv；Population M is set；
Step A2, state vector is according to a preliminary estimate when kth time charge and discharge cycles: according to formula:
It realizes and state vector when kth time recycles is obtained by the final estimated value recursion of state vector when kth -1 time circulation
It is worth according to a preliminary estimate,
In formula,It is kth time circulation to state variable XiEstimated value；wiIt (k) is state variable XiSystematic procedure
Noise, obey mean value be 0, variance σw,iGaussian Profile；
Step A3, particle sampler when kth time charge and discharge cycles: state vector when each circulation corresponds to M particle, root
According to formula 1, particle when kth time charge and discharge cycles is determined；
Step A4, it calculates importance weight: bringing M particle into observational equation in formula 6 respectively, obtain seeing capacity
M estimated value of measurementWherein,Xp,jIt (k) is jth
A particle,For the corresponding observed quantity estimated value of j-th of particle, 1≤j≤M,
According to M estimated valueWith actual measurement observed quantity Q (k) error, according to formula:
The importance weight of different particles is obtained,
In formula, WjIt (k) is the importance weight of j-th of particle, σvIt is σ for variancevGaussian Profile；
Step A5, weight normalizes: according to formula:
By the importance weight of each particle divided by the sum of all particle importance weights, each particle importance power is realized
The normalization of value,
In formula,For the importance weight after normalization；
Step A6, particle resampling: resampling is carried out to particle, returns each particle equal to it by the probability of resampling
One importance weight changed；
Step A7, particle updates: according to formula:
Using the average value of each dimension of M particle after resampling as the final estimated value of corresponding states vector,
Training stage, for each charge and discharge cycles, repeat the above steps A2 to step A7, makes to the estimated value of state vector more and more
Close to its true value.
In present embodiment, for convenience, indicate that the index of observation data sequence, Q indicate observed quantity, w table using k
Show that process noise, v indicate observation noise, σwIndicate process-noise variance, σvIndicate observation noise variance.Particle sampler: according to shape
State equation, from kth -1 time M state variable particle, M state variable particle of recursion kth time.
It calculates importance weight: bringing the corresponding quantity of state of each particle into observational equation, obtain the estimation to observed quantity
ValueAccording to the error respectively with actual measurement observed quantity Q, the importance weight of different particles is obtained.
The importance weight of each particle realizes each particle importance weight divided by the sum of all particle importance weights
Normalization.
Specific embodiment 7: present embodiment is to a kind of particle filter described in specific embodiment one and mechanism mould
The service life of secondary cell prediction technique that type combines is described further, and in present embodiment, in step 3, predicts the tool of process
Body process are as follows:
Step B1, state equation updates: in the training stage finally, being become according to the variation of historic state vector estimated value
Gesture obtains the recurrence polynomial equation that they change with charge and discharge cycles number k, adds systematic procedure error, is updated
State equation afterwards, updated state equation are as follows:
X (k)=fregression(k)+w (k), w~N (0, σw) (formula 11),
In formula, fregression(k) indicate that regression equation of the quantity of state about k, w (k) are systematic procedure noise；
Step B2, observed quantity updates: by estimating for the state vector X (k) after the kth obtained in step B1 time charge and discharge cycles
Evaluation substitutes into the estimated value in formula 1 as initial value, obtains M particle, which is substituted into formula 6, obtains M
The estimated value of observation, capacity is pre- when using the median of the estimated value of M observation as to the following kth time charge and discharge cycles
Measured value；
Step B3, predicting residual useful life: most by the predicted value of capacity when kth time charge and discharge cycles and preset battery
Low capacity is compared, with the increase of charge and discharge cycles number, when predicting that capacity starts to be less than setting capacity, the reality of battery
The difference of cycle-index k corresponding to the observed quantity of border and cycle-index used in the training stage in step 2 is that battery is available surplus
Remaining cycle-index, to obtain secondary cell remaining life.
Claims (7)
1. the service life of secondary cell prediction technique that a kind of particle filter is combined with mechanism model, which is characterized in that it include with
Lower content:
Step 1: the mechanism model of building secondary cell, the mechanism model of the secondary cell can simulate any current condition
When battery the curve that changes over time of charging/discharging voltage；
Step 2: the training stage: the secondary cell in step 1 is subjected to aging for a period of time under normal use operating condition, every
Every fixed charge and discharge cycles number using the charging and discharging curve in dynamic operation condition off-line measurement secondary cell ageing process, at this time
The voltage of acquisition is practical secondary cell output voltage U,
Same dynamic operation condition electric current is inputted to the emulation of the mechanism model of secondary cell, is exported with model emulationIt goes to simulate secondary
Actual output voltage U of the battery in the different ageing steps, using genetic algorithm or least square method, according to objective function realization pair
The identification of secondary cell model parameter collection P, the multiple parameters collection P for secondary cell each ageing step that identification is obtained is as instruction
Practice data,
Select L mechanism model parameter relevant to ageing process as state vector X from the multiple parameters collection P that uses of training,
Wherein, L is positive integer, using the battery capacity Q under actual load current conditions as observed quantity, under same load current conditions
Battery mechanism model emulation and conversion capacity estimation valueProcess as observational equation, make aging using particle filter algorithm
The estimated value of the state vector in each stage in journeyClose to true value X；
Step 3: prediction process: using the state vector estimated value sequence passed through in particle filter algorithm training process in step 2
Column, with the method for polynomial regression, obtain regression equation of the state vector X about cycle-index k, in this, as new state side
Journey obtains the estimated value of state vector by new state equation when k is some following charge and discharge cycles numberGeneration
Enter equation:
In formula, k is cycle-index, XP,i,j(k) i-th of component for indicating j-th of particle, obeying mean value isVariance is
σw,iNormal distribution, 1≤i≤L, 1≤j≤M, wi(k) the systematic procedure noise for being state variable Xi,
The multiple particles for meeting Gaussian Profile are obtained, multiple particles are substituted into the observational equation in step 2, it is more with what is obtained
The median of a observed quantity estimated value, as the predicted value to the following battery capacity, when prediction capacity reaches preset electricity
When tankage lower limit, the difference of corresponding cycle-index and cycle-index used in the training stage in step 2 is that battery is available
Cycles left number, to realize the prediction to secondary cell remaining life.
2. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, in step 1, the mechanism model of secondary cell are as follows:
U (t)=f [I (t), P (k)] (formula 2),
In formula, I (t) is to constant current, and f is Function Mapping, and P (k) is secondary cell model parameter collection, and k is charge and discharge cycles time
Number, parameter set P change with the increase of charge and discharge number k, and U (t) is the external voltage-measurable of secondary cell.
3. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, in step 2, objective function are as follows:
In formula, I (t) is to constant current, and P is parameter set to be identified, and S is the search space of parameter set, N for the voltage that is taken at any time
Between data points on change curve,For mechanism model simulation data voltage, U is actual battery output voltage.
4. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, in step 2, state vector X are as follows:
In formula, X1~XLFor ageing-related in model parameter collection P L parameter.
5. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, in step 2, the equation of battery capacity Q are as follows:
Q (k)=q [I (t), P (k)] (formula 5),
In formula, I (t) is electric current used by measurement capacity, and P (k) is secondary cell model parameter collection, and q [] indicates that basis is put
Electric curve and discharge time calculate the process of discharge capacity；
In step 2, observational equation are as follows:
In formula,For the estimated value of the observed quantity under specific charge and discharge operating condition I (t), k is charge and discharge cycles number, and v is that observation is made an uproar
Sound, obey mean value be 0, variance σvGaussian Profile.
6. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, making the estimated value of the state vector in each stage in ageing process using particle filter algorithm in step 2
Close to the detailed process of true value X are as follows:
Step A1, particle initialization is carried out using particle filter algorithm: setting state vector XiProcess-noise variance σw,i, 1≤i
≤L；Observation noise variances sigma is setv；Population M is set；
Step A2, state vector is according to a preliminary estimate when kth time charge and discharge cycles: according to formula:
Realize the preliminary of state vector when obtaining kth time circulation by the final estimated value recursion of state vector when kth -1 time circulation
Estimated value,
In formula,It is kth time circulation to state variable XiEstimated value；wiIt (k) is state variable XiSystematic procedure noise,
Obey mean value be 0, variance σw,iGaussian Profile；
Step A3, particle sampler when kth time charge and discharge cycles: state vector when each circulation corresponds to M particle, according to public affairs
Formula 1 determines particle when kth time charge and discharge cycles；
Step A4, it calculates importance weight: bringing M particle into observational equation in formula 6 respectively, obtain to capacity observed quantity
M estimated valueWherein,Xp,jIt (k) is j-th
Son,For the corresponding observed quantity estimated value of j-th of particle, 1≤j≤M,
According to M estimated valueWith actual measurement observed quantity Q (k) error, according to formula:
The importance weight of different particles is obtained,
In formula, WjIt (k) is the importance weight of j-th of particle, variance σvGaussian Profile；
Step A5, weight normalizes: according to formula:
By the importance weight of each particle divided by the sum of all particle importance weights, each particle importance weight is realized
Normalization,
In formula,For the importance weight after normalization；
Step A6, particle resampling: carrying out resampling to particle, makes each particle by the probability of resampling equal to its normalization
Importance weight；
Step A7, particle updates: according to formula:
Using the average value of each dimension of M particle after resampling as the final estimated value of corresponding states vector, in training
In the stage, for each charge and discharge cycles, repeat the above steps A2 to step A7, keeps the estimated value to state vector more and more close
Its true value.
7. the service life of secondary cell prediction technique that a kind of particle filter according to claim 1 is combined with mechanism model,
It is characterized in that, predicting the detailed process of process in step 3 are as follows:
Step B1, state equation updates: in the training stage finally, according to the variation tendency of historic state vector estimated value, obtaining
The recurrence polynomial equation changed to them with charge and discharge cycles number k, adds systematic procedure error, obtains updated shape
State equation, updated state equation are as follows:
X (k)=fregression(k)+w (k), w~N (0, σw) (formula 11),
In formula, fregression(k) indicate that regression equation of the quantity of state about k, w (k) are systematic procedure noise；
Step B2, observed quantity updates: by the estimated value of the state vector X (k) after the kth obtained in step B1 time charge and discharge cycles
As initial value, which is substituted into formula 1, obtains M particle, which is substituted into formula 6, obtains M observation
The estimated value of value, the prediction of capacity when using the median of the estimated value of M observation as to the following kth time charge and discharge cycles
Value；
Step B3, predicting residual useful life: the predicted value of capacity when kth time charge and discharge cycles and preset battery minimum are held
Amount is compared, with the increase of charge and discharge cycles number, when predicting that capacity starts to be less than setting capacity, the practical sight of battery
The difference of the corresponding cycle-index k of measurement and cycle-index used in the training stage in step 2 is that the available residue of battery is followed
Ring number, to obtain secondary cell remaining life.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610363499.1A CN106055775B (en) | 2016-05-27 | 2016-05-27 | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610363499.1A CN106055775B (en) | 2016-05-27 | 2016-05-27 | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106055775A CN106055775A (en) | 2016-10-26 |
| CN106055775B true CN106055775B (en) | 2018-12-11 |
Family
ID=57175367
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610363499.1A Active CN106055775B (en) | 2016-05-27 | 2016-05-27 | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106055775B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019018974A1 (en) * | 2017-07-24 | 2019-01-31 | 罗伯特·博世有限公司 | Method and system for performing modeling and estimation of battery capacity |
| CN109543317B (en) * | 2018-04-28 | 2022-11-01 | 北京航空航天大学 | Method and device for predicting remaining service life of PEMFC |
| CN109146115A (en) * | 2018-06-11 | 2019-01-04 | 广州市香港科大霍英东研究院 | Battery life predicting method, system and device based on model migration |
| CN108761347B (en) * | 2018-06-29 | 2021-01-15 | 联想(北京)有限公司 | Processing method and device and electronic equipment |
| CN110764003A (en) * | 2018-07-10 | 2020-02-07 | 天津工业大学 | Lithium battery state of charge estimation method, device and system |
| CN112084459A (en) * | 2019-06-12 | 2020-12-15 | 电计贸易(上海)有限公司 | Method and device for predicting battery charge-discharge cycle life, electronic terminal and storage medium |
| CN110580382B (en) * | 2019-08-16 | 2021-10-08 | 华中科技大学 | Capacitance prediction method for metallized film capacitor |
| CN110687450B (en) * | 2019-08-28 | 2022-10-25 | 武汉科技大学 | Lithium battery residual life prediction method based on phase space reconstruction and particle filtering |
| CN112130086B (en) * | 2020-05-29 | 2023-10-31 | 国家电网有限公司 | Method and system for predicting remaining life of power battery |
| CN114779088B (en) * | 2022-04-20 | 2023-05-12 | 哈尔滨工业大学 | Battery remaining service life prediction method based on maximum expected-unscented particle filtering |
| CN115114878B (en) * | 2022-07-26 | 2022-12-27 | 中国长江三峡集团有限公司 | Method and device for online prediction of battery life of energy storage power station and storage medium |
| CN117347888B (en) * | 2023-12-05 | 2024-02-13 | 深圳市易检车服科技有限公司 | Battery equalizer load measuring method and device, computer equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102798823A (en) * | 2012-06-15 | 2012-11-28 | 哈尔滨工业大学 | Gaussian process regression-based method for predicting state of health (SOH) of lithium batteries |
| CN102968573A (en) * | 2012-12-14 | 2013-03-13 | 哈尔滨工业大学 | Online lithium ion battery residual life predicting method based on relevance vector regression |
| CN105445671A (en) * | 2015-12-29 | 2016-03-30 | 北京航天测控技术有限公司 | Lithium ion battery service life prediction method based on traceless particle filtering |
-
2016
- 2016-05-27 CN CN201610363499.1A patent/CN106055775B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102798823A (en) * | 2012-06-15 | 2012-11-28 | 哈尔滨工业大学 | Gaussian process regression-based method for predicting state of health (SOH) of lithium batteries |
| CN102968573A (en) * | 2012-12-14 | 2013-03-13 | 哈尔滨工业大学 | Online lithium ion battery residual life predicting method based on relevance vector regression |
| CN105445671A (en) * | 2015-12-29 | 2016-03-30 | 北京航天测控技术有限公司 | Lithium ion battery service life prediction method based on traceless particle filtering |
Non-Patent Citations (3)
| Title |
|---|
| 基于数据驱动的卫星锂离子电池寿命预测方法;艾力 等;《计算机测量与控制》;20150430;第23卷(第4期);第1262-1265,1272页 * |
| 蓄电池SOC估算方法综述;季迎旭 等;《电测与仪表》;20140225;第51卷(第4期);第18-22页 * |
| 锂离子电池健康评估和寿命预测综述;刘大同 等;《仪器仪表学报》;20150131;第36卷(第1期);第1-16页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106055775A (en) | 2016-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106055775B (en) | A kind of service life of secondary cell prediction technique that particle filter is combined with mechanism model | |
| CN110568359B (en) | Lithium battery residual life prediction method | |
| CN106842045B (en) | Battery multi-model fusion modeling method and battery management system based on self-adaptive weight method | |
| Lujano-Rojas et al. | Operating conditions of lead-acid batteries in the optimization of hybrid energy systems and microgrids | |
| CN110488194B (en) | Lithium battery SOC estimation method and system based on electrochemical impedance model | |
| CN113064093B (en) | Method and system for jointly estimating state of charge and state of health of energy storage battery | |
| Shi et al. | Battery health management using physics-informed machine learning: Online degradation modeling and remaining useful life prediction | |
| CN103389471B (en) | A kind of based on the cycle life of lithium ion battery indirect predictions method of GPR with indeterminacy section | |
| Li et al. | On-line estimation method of lithium-ion battery health status based on PSO-SVM | |
| JP5936711B2 (en) | Storage device life prediction apparatus and storage device life prediction method | |
| CN105203963B (en) | A kind of method of estimation of the state-of-charge based on open-circuit voltage hysteretic characteristic | |
| CN113805064B (en) | Lithium ion battery pack health state prediction method based on deep learning | |
| Zhang et al. | Aging performance characterization and state-of-health assessment of retired lithium-ion battery modules | |
| CN111426957B (en) | SOC estimation optimization method for power battery under simulated vehicle working condition | |
| CN113702843B (en) | Lithium battery parameter identification and SOC estimation method based on suburb optimization algorithm | |
| CN105334462A (en) | Online estimation method for battery capacity loss | |
| CN109932656A (en) | A kind of service life of lithium battery estimation method based on IMM-UPF | |
| CN104569844A (en) | Valve control seal type lead-acid storage battery health condition monitoring method | |
| CN110082682A (en) | A kind of lithium battery charge state estimation method | |
| CN110673037B (en) | Battery SOC estimation method and system based on improved simulated annealing algorithm | |
| CN103424712A (en) | Method for measuring residual capacity of battery in online manner on basis of particle swarm optimization | |
| CN106597288A (en) | Power supply SOC estimation method | |
| Liu et al. | Battery degradation model and multiple-indicators based lifetime estimator for energy storage system design and operation: Experimental analyses of cycling-induced aging | |
| CN115587527A (en) | Battery life prediction method, system, terminal device and computer readable medium | |
| CN204030697U (en) | Based on the battery management system of dynamic SOC estimating system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20191025 Address after: 519175 sancunpian, Fushan Industrial Zone, Doumen District, Zhuhai City, Guangdong Province (plant 4, plant 6) Patentee after: Zhuhai Zhongli New Energy Technology Co., Ltd. Address before: 150001 Harbin, Nangang, West District, large straight street, No. 92 Patentee before: Harbin Institute of Technology |