CN106896324A - A kind of SOC methods of estimation - Google Patents
A kind of SOC methods of estimation Download PDFInfo
- Publication number
- CN106896324A CN106896324A CN201710053972.0A CN201710053972A CN106896324A CN 106896324 A CN106896324 A CN 106896324A CN 201710053972 A CN201710053972 A CN 201710053972A CN 106896324 A CN106896324 A CN 106896324A
- Authority
- CN
- China
- Prior art keywords
- estimation
- time
- state
- soc
- equation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention discloses a kind of lithium battery SOC methods of estimation, the output valve and model of each measurement being estimated, the weighted sum of the output valve residual error of residual error and each state sigma points estimation of the output valve for obtaining estimates the noise covariance at current time as new breath, it is allowed to be updated with the time, Real-time Feedback, changeless noise covariance originally can be allowed to be updated with the time, Real-time Feedback, so as to improve precision, there is preferable robustness to initial value, right value can be faster converged to during recurrence calculation, and estimation precision is high, dynamic property is good, amount of calculation is small, it is convenient to realize.
Description
Technical field
The present invention relates to electrokinetic cell management domain, more particularly to a kind of SOC methods of estimation.
Background technology
In recent years, with air quality go from bad to worse and petroleum resources gradually deficient, new-energy automobile, especially
Pure electric automobile turns into the exploitation focus of major motor corporations of the world today.Power battery pack as electric automobile crucial portion
Part, electrokinetic cell SOC is used to the dump energy of direct reaction battery, is that whole-control system formulates optimal energy management strategy
Important evidence, the accurate estimation of power electric ground SOC value for improve cell safety reliability, improve energy content of battery utilization rate,
Extension battery life is significant.At present, conventional SOC methods of estimation mainly have open circuit voltage method, current integration method, card
Kalman Filtering method and neural network etc..
Open circuit voltage method, according to OCV-SOC relations, is used alone to the parked state suitable for electric automobile, it is impossible to
Line, dynamic estimation.Generally, open circuit voltage method is used to be provided for other methods of estimation the initial value of SOC.
The general principle of ampere-hour integration method:Current integration method has the advantages that low cost, measures convenient, and shortcoming is needs
SOC initial values are obtained by other methods;Current measurement precision has decisive influence to SOC estimated accuracies;Integral process
Accumulated error cannot be eliminated.
Neural network has good non-linear mapping capability, and the nonlinear characteristic of electrokinetic cell can be preferable in theory
Mapped by neutral net, but it needs substantial amounts of data to be trained, and using complexity, training data and training method are to estimating
The influence of precision is larger.
Kalman filtering method (KF, KalmanFilter) is applied in linear system, and core concept is to dynamical system
State makes the optimal estimation in lowest mean square meaning, and the advantage of Kalman filtering is stronger error correcting capabilities, and deficiency exists
Higher is relied on to the accuracy of battery model in estimated accuracy.Going out EKF derived from Organization of African Unity's sexual system in recent years
With Unscented kalman filtering scheduling algorithm.
Spreading kalman filter is draped over one's shoulders (EKF) and carries out first-order linear to the Taylor expansions of nonlinear function and blocks, and is ignored
Remaining high-order top, so as to nonlinear problem be converted into linearly, can be applied to nonlinear system by Kalman's linear filtering algorithm
In system.So, nonlinear problem is solved.But this kind of method also brings two shortcomings, one is working as strong nonlinearity
When EKF run counter to local linear it is assumed that when ignored higher order term brings big error in Taylor expansions, EKF algorithms may
Filtering divergence can be made;Further, since EKF needs to use Jacobian matrix in linearization process, its cumbersome calculating process causes
The method realizes relative difficulty.
Unscented kalman filtering (UKF) is a kind of new filtering algorithm for estimating.UKF is abandoned based on UT conversion
The traditional method linearized to nonlinear function, using Kalman's linear filtering framework, for one-step prediction equation, uses
Become the non-linear transmission for bringing treatment average and covariance without mark (UT).UKF is not linearized and is ignored higher order term, therefore non-linear
The computational accuracy of distribution statisticses amount is higher.But traditional Unscented kalman filtering algorithm covariance regards constant as, it is impossible to meet
The characteristic of noise real-time update, so as to generate certain influence to precision.
The content of the invention
The purpose of the present invention is improved on the basis of common UKF algorithms, by the output valve and model of each measurement
The weighted sum of the output valve residual error of residual error and each state sigma points estimation of the output valve that estimation is obtained is estimated as new breath
The noise covariance at current time, allows it to be updated with the time, Real-time Feedback, so as to improve precision.
To achieve the above object, the technical scheme of present invention offer is:A kind of SOC methods of estimation, comprise the following steps,
S1, set up SOC equation of the lithium battery group in t:
WhereinkiIt is charge-discharge magnification penalty coefficient, ktIt is temperature compensation coefficient, kcIt is for cycle-index is compensated
Number, CNIt is the actual active volume of battery;
S2, equivalent circuit modeling is carried out to lithium battery, equivalent-circuit model is established as using Order RC circuit, the equivalent electric
Road model includes the voltage source Voc, the internal resistance R that are sequentially connected in series and the different RC parallel connection loops of two time constants;
S3, the state equation for setting up SOC:
Wherein:
S4, according to step S3 set up state equation, update state equation estimate time be
S5, update the mean square deviation time be
S6, prior estimate system are output as
S7, calculating filtering gain matrix;
S8, the state equation updated according to step S4 estimate time, the system output prior estimate of step S6, step S7
Filtering gain matrix are estimating optimal state
S9, the mean square deviation time updated according to step S5, the filtering gain matrix of step S7 are estimating mean square error
S10, process noise and measurement noise renewal equation are set up,
Wherein μkAnd yK | k-1, iIt is respectively to measure the measurement output quantity that each sigma points estimation of residual sum of output quantity is obtained
Residual error.
Specifically, the method for building up of step S4 state equations is:
S41, determine that original state is
S42, the original state in S41 is augmented:
WhereinQ, R are covariance matrixes, are right
The matrix of title, and be the variance in each dimension on diagonal;
S43, in S42 extended mode calculate average:
Variance is calculated to the extended mode in S42:
S44,2L+1 sampled point of selection, wherein Sample={ zi, XK-1, i, i=0,1,2 ..., 2L+1;XK-1, iFor selected
Particle, ziIt is corresponding weighted value.
Specifically, step S44 particle points are chosen as follows:
Corresponding weight coefficient is:
Wherein λ is proportionality coefficient, meets λ=α2(L+t)-L, z(m)、z(c)It is respectively that particle point average is corresponding with variance
Weighted value;AndRepresent (L+ λ) PX, k-1On Square-Rooting Matrices i-th row.
Specifically, process noise and measurement noise renewal equation that step S10 sets up, by the residual of the output valve of each measurement
Difference μkThe residual error y of the output valve obtained with model estimationkAnd the output valve residual error y of each state sigma points estimationK | k-1, iWeighting
Noise covariance with current time is estimated as new breath, allows it to be updated with the time, Real-time Feedback, so as to improve precision.
The present invention has the advantage that and allows changeless noise covariance originally to be updated with the time, anti-in real time
Feedback, so as to improve precision, has preferable robustness to initial value, can be converged to faster during recurrence calculation correct
Value, and estimation precision is high, dynamic property is good, and amount of calculation is small, convenient to realize.
Brief description of the drawings
Fig. 1 is equivalent-circuit model schematic diagram of the present invention.
Specific embodiment
1 introduce specific embodiment of the invention referring to the drawings.
The general type of nonlinear discrete time system state equation and observational equation is
ωkBe system noise, vkIt is observation noise, for UKF, its iterative equation is one group based on certain selected and adopts
Come what is carried out, the basis for selecting of sampled point is its average and variance is kept with the expectation average and variance of state variable to sampling point
Unanimously, then these points are transmitted by nonlinear system model, a series of prediction point group are produced, finally by these
The average and variance of point group are corrected, just it is estimated that desired average and variance, before UKF iteration, first become state
Amount expands to the superposition of reset condition, process noise and measurement noise three.For lithium battery, can be obtained in t by current integration method
The SOC of moment battery pack is:
WhereinkiIt is charge-discharge magnification penalty coefficient, ktIt is temperature compensation coefficient, kcIt is for cycle-index is compensated
Number, CNIt is the actual active volume of battery.
General type according to nonlinear discrete time system state equation and observational equation andObtain state equation:
Uk=Ek-IkR-US, k-UP, k+v(k)
=F (SOCk)-IkR-US, k-UP, k+v(k)
Wherein:
1 circuit model with reference to the accompanying drawings:
xk=[SOCk, US, k, UP, k] it is the reset condition of system;ykIt is original output, the U in corresponding circuits modelk;ukFor
Controlled quentity controlled variable, the I in corresponding circuits modelK, and make ψ=[y1, y2..., yk], then carry out Unscented kalman filtering computing.
First carry out the renewal of state estimation time:Based on last moment state optimization estimate be expanded state average and
Variance, selects (2L+1) individual sampled point accordingly, and sampled point finally is entered into line translation and completion status prediction by state equation.
Initialization, determines original state:
State is augmented:
Wherein, Q, R are covariance matrixes, are symmetrical matrixes, and are the variances in each dimension on diagonal.
Extended mode average:
Extended mode variance:
Choose sampled point:Sample={ zi, XK-1, i, i=0,1,2 ..., 2L+1;Wherein XK-1, iIt is selected particle, ziIt is
Corresponding weighted value.Particle point is chosen as follows:
Corresponding weight coefficient is:
λ is ratio
Example coefficient, meets:λ=α2(L+t)-L, z(m)、z(c)It is respectively the particle point average weighted value corresponding with variance;AndRepresent (L+ λ) PX, k-1On Square-Rooting Matrices i-th row;Parameter t meets t >=0 to ensure that variance matrix is
Positive definite, typically gives tacit consent to t=0;α controls particle distribution distance, and meets 10-2≤ α≤1, takes α=1 herein;β is in reduction higher order term
Error, to taking β=2 so that normal distribution is optimal.Analytical sampling pointCan be divided into againWithThree parts, according to
This time for carrying out state estimation is updated to:
The mean square error time updates:
System exports prior estimate:
Filtering gain matrix are calculated:
State optimization is estimated:
Mean square error is estimated:
Because process noise and measurement noise are all time-varying, in order to allow noise covariance real-time update:
Wherein μkAnd yK | k-1, iIt is respectively to measure the measurement output quantity that each sigma points estimation of residual sum of output quantity is obtained
Residual error.Can implementation process noise and measurement noise real-time update.So far ART network Kalman filtering is set up and is completed.
Traditional Unscented kalman filtering algorithm covariance regards constant as, it is impossible to meet the characteristic of noise real-time update, from
And certain influence is generated to precision, to eliminate this influence, the present invention is improved on the basis of common UKF algorithms, walks
Process noise and measurement noise renewal equation that rapid S10 sets up, by the residual error μ of the output valve of each measurementkEstimate with model
The residual error y of the output valve for arrivingkAnd the output valve residual error y of each state sigma points estimationK | k1, iWeighted sum estimate as new breath
The noise covariance at current time, allows it to be updated with the time, Real-time Feedback, so as to improve precision.
The above is not imposed any restrictions to technical scope of the invention, all according to the technology of the present invention essence, to the above
Embodiment any modification, equivalent variations and the modification made, still fall within the range of technical scheme.
Claims (4)
1. a kind of SOC methods of estimation, it is characterised in that comprise the following steps,
S1, set up SOC equation of the lithium battery group in t:
WhereinkiIt is charge-discharge magnification penalty coefficient, ktIt is temperature compensation coefficient, kcIt is cycle-index penalty coefficient, CN
It is the actual active volume of battery;
S2, equivalent circuit modeling is carried out to lithium battery, equivalent-circuit model is established as using Order RC circuit, the equivalent circuit mould
Type includes the voltage source Voc, the internal resistance R that are sequentially connected in series and the different RC parallel connection loops of two time constants;
S3, the state equation for setting up SOC:
Wherein:
S4, according to step S3 set up state equation, update state equation estimate time be
S5, update the mean square deviation time be
S6, prior estimate system are output as
S7, calculating filtering gain matrix;
S8, the state equation updated according to step S4 estimate time, the system output prior estimate of step S6, the filtering of step S7
Gain matrix is estimating optimal state
S9, the mean square deviation time updated according to step S5, the filtering gain matrix of step S7 are estimating mean square error
S10, process noise and measurement noise renewal equation are set up,
Wherein, μkAnd yK | k-1, iIt is respectively to measure the measurement output quantity that residual sum each state sigma points estimation of output quantity is obtained
Residual error.
2. a kind of SOC methods of estimation according to claim 1, it is characterised in that the method for building up of step S4 state equations
For:
S41, determine that original state is
S42, the original state in S41 is augmented:
WhereinQ, R are covariance matrixes, are symmetrical squares
Battle array, and be the variance in each dimension on diagonal;
S43, in S42 extended mode calculate average:
Variance is calculated to the extended mode in S42:
S44,2L+1 sampled point of selection, wherein Sample={ zi, XK-1, i, i=0,1,2 ..., 2L+1;XK-1, iIt is selected grain
Son, ziIt is corresponding weighted value.
3. a kind of SOC methods of estimation according to claim 2, it is characterised in that step S44 particle points are selected as follows
Take:
Corresponding weight coefficient is:
Wherein λ is proportionality coefficient, meets λ=α2(L+t)-L, z(m)、z(c)It is respectively the weighting corresponding with variance of particle point average
Value;AndRepresent (L+ λ) PX, k-1On Square-Rooting Matrices i-th row.
4. a kind of SOC methods of estimation according to claim 1, it is characterised in that process noise and survey that step S10 sets up
Amount noise renewal equation, by the residual error μ of the output valve of each measurementkThe residual error y of the output valve obtained with model estimationkAnd each shape
The output valve residual error y of state sigma points estimationK | k-1, iWeighted sum the noise covariance at current time is estimated as new breath, allow
It updates with the time, Real-time Feedback, so as to improve precision.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053972.0A CN106896324A (en) | 2017-01-20 | 2017-01-20 | A kind of SOC methods of estimation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710053972.0A CN106896324A (en) | 2017-01-20 | 2017-01-20 | A kind of SOC methods of estimation |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106896324A true CN106896324A (en) | 2017-06-27 |
Family
ID=59199052
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710053972.0A Pending CN106896324A (en) | 2017-01-20 | 2017-01-20 | A kind of SOC methods of estimation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106896324A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107741569A (en) * | 2017-11-16 | 2018-02-27 | 温州大学 | A kind of evaluation method of the lithium battery charge state based on segment extension Kalman filtering |
CN109256844A (en) * | 2018-11-01 | 2019-01-22 | 三峡大学 | A kind of electric car wireless charging circuit and charge control method |
CN109298351A (en) * | 2018-09-30 | 2019-02-01 | 清华大学深圳研究生院 | A kind of new energy on-vehicle battery remaining life estimation method based on model learning |
CN109606200A (en) * | 2018-12-19 | 2019-04-12 | 江苏科达车业有限公司 | A kind of new energy car battery management system |
CN109856556A (en) * | 2019-03-21 | 2019-06-07 | 南京工程学院 | A kind of power battery SOC estimation method |
CN110888063A (en) * | 2019-12-02 | 2020-03-17 | 上海国际港务(集团)股份有限公司 | SOC estimation method based on design of port machine tire crane lithium iron phosphate battery parallel system |
CN111044906A (en) * | 2019-12-10 | 2020-04-21 | 深圳市鹏诚新能源科技有限公司 | Lithium ion battery energy state estimation method based on maximum likelihood criterion |
CN112255545A (en) * | 2019-07-05 | 2021-01-22 | 西南科技大学 | Lithium battery SOC estimation model based on square root extended Kalman filter |
US11231466B2 (en) | 2017-09-07 | 2022-01-25 | Lg Energy Solution, Ltd. | Apparatus and method for estimating a state of charge of a battery |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101615794A (en) * | 2009-08-05 | 2009-12-30 | 河海大学 | Electrical Power System Dynamic method for estimating state based on no mark transformation Kalman filter |
CN103675706A (en) * | 2013-12-13 | 2014-03-26 | 桂林电子科技大学 | Power battery electric charge quantity estimation method |
CN104573294A (en) * | 2013-10-15 | 2015-04-29 | 胡志坤 | Self-adaptive kalman filter estimation algorithm for power battery |
CN105093122A (en) * | 2015-07-09 | 2015-11-25 | 宁波飞拓电器有限公司 | Strong-tracking self-adaptive-SQKF-based SOC estimation method of emergency lamp battery |
-
2017
- 2017-01-20 CN CN201710053972.0A patent/CN106896324A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101615794A (en) * | 2009-08-05 | 2009-12-30 | 河海大学 | Electrical Power System Dynamic method for estimating state based on no mark transformation Kalman filter |
CN104573294A (en) * | 2013-10-15 | 2015-04-29 | 胡志坤 | Self-adaptive kalman filter estimation algorithm for power battery |
CN103675706A (en) * | 2013-12-13 | 2014-03-26 | 桂林电子科技大学 | Power battery electric charge quantity estimation method |
CN105093122A (en) * | 2015-07-09 | 2015-11-25 | 宁波飞拓电器有限公司 | Strong-tracking self-adaptive-SQKF-based SOC estimation method of emergency lamp battery |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11231466B2 (en) | 2017-09-07 | 2022-01-25 | Lg Energy Solution, Ltd. | Apparatus and method for estimating a state of charge of a battery |
CN107741569A (en) * | 2017-11-16 | 2018-02-27 | 温州大学 | A kind of evaluation method of the lithium battery charge state based on segment extension Kalman filtering |
CN109298351A (en) * | 2018-09-30 | 2019-02-01 | 清华大学深圳研究生院 | A kind of new energy on-vehicle battery remaining life estimation method based on model learning |
CN109256844A (en) * | 2018-11-01 | 2019-01-22 | 三峡大学 | A kind of electric car wireless charging circuit and charge control method |
CN109606200A (en) * | 2018-12-19 | 2019-04-12 | 江苏科达车业有限公司 | A kind of new energy car battery management system |
CN109606200B (en) * | 2018-12-19 | 2020-12-22 | 江苏科达车业有限公司 | New energy automobile battery management system |
CN109856556A (en) * | 2019-03-21 | 2019-06-07 | 南京工程学院 | A kind of power battery SOC estimation method |
CN112255545A (en) * | 2019-07-05 | 2021-01-22 | 西南科技大学 | Lithium battery SOC estimation model based on square root extended Kalman filter |
CN110888063A (en) * | 2019-12-02 | 2020-03-17 | 上海国际港务(集团)股份有限公司 | SOC estimation method based on design of port machine tire crane lithium iron phosphate battery parallel system |
CN111044906A (en) * | 2019-12-10 | 2020-04-21 | 深圳市鹏诚新能源科技有限公司 | Lithium ion battery energy state estimation method based on maximum likelihood criterion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106896324A (en) | A kind of SOC methods of estimation | |
CN105116343B (en) | The electrokinetic cell state of charge method of estimation and system of least square method supporting vector machine | |
CN108594135A (en) | A kind of SOC estimation method for the control of lithium battery balance charge/discharge | |
CN110659722B (en) | Electric vehicle lithium ion battery health state estimation method based on AdaBoost-CBP neural network | |
CN110286332B (en) | Electric vehicle power battery SOC estimation method based on multiple innovation theory | |
CN107064811A (en) | A kind of lithium battery SOC On-line Estimation methods | |
He et al. | State-of-charge estimation of lithium ion batteries based on adaptive iterative extended Kalman filter | |
CN107367699A (en) | A kind of lithium battery SOC estimation new methods based on fractional model | |
CN107436971B (en) | Improved Latin hypercube sampling method suitable for non-positive stereotype correlation control | |
CN111044916B (en) | Method for jointly estimating parameters and whole-region SOC of power battery | |
CN103983920B (en) | A kind of method of the model of the electrokinetic cell setting up electric vehicle | |
CN108089135A (en) | A kind of battery status forecasting system and its implementation based on limit learning model | |
CN107831448A (en) | A kind of state-of-charge method of estimation of parallel connection type battery system | |
CN106291376A (en) | Lithium battery SOC method of estimation based on supporting vector machine model and Kalman filtering | |
CN114184962B (en) | Multi-algorithm fusion lithium ion battery SOC and SOH joint estimation method | |
CN112858920B (en) | SOC estimation method of all-vanadium redox flow battery fusion model based on adaptive unscented Kalman filtering | |
CN106443496A (en) | Battery charge state estimation method with improved noise estimator | |
CN114660464A (en) | Lithium ion battery state of charge estimation method | |
CN106154168A (en) | The method for estimating charge state of power cell of data-driven | |
CN113608121A (en) | Lithium battery SOC estimation method based on fuzzy fractional order unscented Kalman filtering | |
CN116125278A (en) | Lithium battery SOC estimation method and system based on LSTM-EKF algorithm | |
CN115656848A (en) | Lithium battery SOC estimation method based on capacity correction | |
CN110441691A (en) | It is a kind of based on the SOC estimation method for simplifying particle Unscented transform | |
CN114217234B (en) | IDE-ASRCKF-based lithium ion battery parameter identification and SOC estimation method | |
CN113419177B (en) | Extended Kalman filtering SOC estimation method based on combination of improved particle swarm algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170627 |
|
RJ01 | Rejection of invention patent application after publication |