CN106772074A - A kind of battery system state-of-charge method of estimation with noise estimator - Google Patents
A kind of battery system state-of-charge method of estimation with noise estimator Download PDFInfo
- Publication number
- CN106772074A CN106772074A CN201611121601.3A CN201611121601A CN106772074A CN 106772074 A CN106772074 A CN 106772074A CN 201611121601 A CN201611121601 A CN 201611121601A CN 106772074 A CN106772074 A CN 106772074A
- Authority
- CN
- China
- Prior art keywords
- state
- noise
- battery system
- estimation
- battery
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
Abstract
The invention discloses a kind of battery system state-of-charge method of estimation with noise estimator, the battery system by battery cell by m go here and there n and m × N-shaped battery system.Methods described is as follows：Cell system space state equation is set up according to battery system equivalent-circuit model, the noise estimation value at k+1 moment is obtained using noise estimatorNoise estimation value is replaced into the noise statisticses information in Unscented kalman filtering method again, and state-of-charge estimation is carried out to battery system using Unscented kalman filtering method with reference to cell system space state equation, obtain the intermediateness amount at k+1 momentAnd as the input quantity of subsequent time noise estimator, circulate recursion with this to obtain battery charge state estimate not in the same time.The present invention using the battery system state-of-charge algorithm for estimating with noise estimator than expanded Kalman filtration algorithm convergence rate faster, robustness it is more preferable.
Description
Technical field
The invention belongs to MW grades of battery energy storage system design and control technology field in intelligent grid, it is related to a kind of band noise
The battery system state-of-charge method of estimation of estimator.
Background technology
Battery system determines its contained electricity as energy stores in battery energy storage system (BESS) and the main carriers of release
Amount number be not only one of major function of battery management system, and can more direct relation BESS and effectively run and control
System.Because battery charge and discharge process is a kind of complicated electrochemical reaction process, and battery electric quantity is difficult directly by sensor
Measure and obtain, battery charge state (State of Charge, SOC) main at present is battery effective electricity and its specified appearance
The ratio of amount characterizes the number of battery electric quantity.
Traditional SOC methods of estimation mainly have ampere-hour method, open circuit voltage method and impedance method etc., occur in that in succession in recent years several
Plant novel advanced algorithm, such as neural network, fuzzy logic method, Kalman filtering method and its innovatory algorithm.Ampere-hour method is simple
And it is easy, it is widely adopted in actual applications, but there is the limitation of error accumulation；For nonlinear system such as BESS
SOC estimates that filtering method (EKF) frequently with spreading kalman carries out battery SOC estimation, but need to calculate Jacobi because EKF has in itself
Matrix, the shortcomings of ignore higher order term, its estimated accuracy still suffers from certain error.Therefore, current many scholars and expert use nothing
Mark Kalman filtering method (UKF) carries out battery SOC estimation.But, because the noise of BESS battery systems in actual motion is united
Meter information (such as system noise, measure noise) is difficult to obtain or inaccurate and there is time variation, causes to carry out SOC using UKF
Its estimated accuracy is still limited during estimation, thus the accuracy that its SOC estimates still requires study improvement.
The content of the invention
It is to provide a kind of battery system state-of-charge method of estimation with noise estimator that the problem that the present invention is solved is,
Battery system being solved when carrying out when SOC estimates using EKF algorithms need to calculate Jacobian matrix and noise statisticses information is unknown or difficulty
Cause the problem that estimated accuracy is not high, convergence rate is slow with acquisition, so as to reach quick, accurate estimation battery system SOC's
Purpose.
The present invention seeks to be achieved through the following technical solutions：
The present invention provides a kind of battery system state-of-charge method of estimation with noise estimator, and the battery system is by battery
Monomer by m go here and there n and m × N-shaped battery system, wherein m, n be the natural number more than 1.
A kind of battery system state-of-charge method of estimation with noise estimator is as follows：First according to known battery system etc.
Effect circuit model (1) sets up cell system space state equation (2), recycles noise estimator (3) to obtain the noise at k+1 moment
Estimate (4), then with the battery system state-of-charge SOC in cell system space state equation (2)b, 2 RC parallel circuits
Terminal voltage as Unscented kalman filtering method UKF (5) state variable, with the input of cell system space state equation (2)
The nonlinear state side of state space equation, output voltage state space equation respectively as Unscented kalman filtering method UKF (5)
Journey f () and measurement equation g (), using noise estimation value (4) the making an uproar as Unscented kalman filtering method UKF (5) at k+1 moment
Sound statistical information, the intermediateness amount (6) at k+1 moment is obtained using Unscented kalman filtering method UKF (5), as subsequent time
The input quantity of noise estimator (3), while exporting the battery charge state SOC at k+1 momentb,k+1, recursion is circulated with this and obtains battery
State-of-charge SOCbEstimate.
The battery system equivalent-circuit model (1) is second order equivalent-circuit model, and model main circuit is by 2 RC electricity in parallel
Road, controlled voltage source Ub0And internal resistance of cell R (SOC)bDeng composition.According to battery system precircuit structure and its charge/discharge operation
Characteristic, the mathematic(al) representation of equivalent-circuit model is：
In formula, a0~a5、b0~b5、c0~c2、d0~d2、e0~e2、f0~f2Model coefficient is, can be by battery measurement number
Obtained according to through fitting；Q0It is battery specified electric quantity；SOC0It is SOC initial values, generally 0~1 constant；Rs、RlBattery is represented respectively
2 resistance and C of RC parallel circuits in monomer models、Cl2 electric capacity of RC parallel circuits in battery cell model are represented respectively；
U0, R represent the open-circuit voltage of battery cell, internal resistance respectively；Ub、IbRespectively battery system terminal voltage and electric current.
The foundation of the cell system space state equation (2) is as follows：1), with battery system SOCbAnd 2 end electricity of RC
Pressure Ubs(t)、UblT () is used as system state variables xk, with Ub、IbRespectively as system measurements variable ykAnd system input variable, root
Setting up cell system space state equation according to equivalent-circuit model is
In formula, Ubs、UblIt is 2 RC parallel circuit terminal voltages, τ1、τ2It is time constant, Δ t is sampling
Cycle, ωkIt is system noise, Δ t is the sampling period, and k is the natural number more than 1；2), according to Kirchhoff's second law, knot
Battery system equivalent-circuit model is closed, can obtain battery system output measurement equation is：[Ub,k]=mU0,k-mRkIb,k/n-Ubs,k-
Ubl,k+υk=gk (xk)+υk=yk, in formula, υkIt is system measurements noise, k is the natural number more than 1.
Unscented kalman filtering algorithm UKF (5) mainly comprise the following steps：1) initialization x averages E () and noise information：
x0=E (x0)、q0、Q0、r0、R0；2) sampled point x is calculatedi,kWith respective weights ω：In formula, λ=α2
(n+h)-n, n are the dimension of state variable；ωm、ωcThe weight of variance and average, operator are represented respectivelyIt is symmetrical matrix
Cholesky is decomposed, and α, β, h are constant；3) the time renewal of state estimation and mean square error：The state estimation time is updated toIn formula,It is state equation noise average；The mean square error time updates
ForQk+1It is state equation noise variance；When system is exported
Between be updated toIn formula, gk-1() is measurement equation, rk+1For measurement equation noise is equal
Value；4) gain matrix is calculated：In formula, Py,kIt is self tuning side
Difference, Rk+1It is state equation noise variance；5) measurement updaue of state estimation and mean square error：State estimation measurement updaue isMean square error measurement updaue is
Described noise estimator (3) is： In formula, Kk+1For gain is joined
Number；ekIt is row difference item, its expression formula isykIt is actual measured value；dk=(1-b)/(1-bk+1), b for forget because
Son, span is 0.95~0.995.
The noise estimation value (4) at described k+1 moment has：The k+1 moment is represented respectively
State equation noise Estimation of Mean value, state equation Noise Variance Estimation value, measurement equation noise Estimation of Mean value, measurement equation
Noise Variance Estimation value.
The intermediateness amount (6) at described k+1 moment has：Pk+1、ek+1, the system that the k+1 moment is represented respectively
State variable estimate, system output variables estimate, noise error covariance, row residual quantity.
Compared with battery system SOC being carried out using expanded Kalman filtration algorithm (EKF) and is estimated, the present invention has with following
The technique effect of benefit：One is whole discharge process, and UKF algorithms of the present invention carry out battery system SOC and estimate than EKF algorithm
Faster, robustness is more preferable for timing convergence rate；Two is used UKF algorithms higher than EKF algorithm estimated accuracies, is especially put
Electric initial stage and latter stage effect become apparent from.
Brief description of the drawings
Fig. 1 is a kind of battery system state-of-charge method of estimation flow chart with noise estimator；
Fig. 2 is m × N-shaped battery system structure schematic diagram；
Fig. 3 is containing 2 battery system equivalent-circuit model figures of RC parallel circuits；
Fig. 4-1~Fig. 4-2 is SOC0Battery constant-current discharge characteristic when=0.8, wherein Fig. 4-1 is SOC0SOC becomes when=0.8
Change situation, Fig. 4-2 is SOC0SOC error conditions when=0.8.
Specific embodiment
With reference to specific example, the present invention is described in further detail, it is described for explanation of the invention without
It is to limit.
According to embodiments of the present invention, as shown in Figure 1, Figure 2, Figure 3 and Figure 4, there is provided the battery system with noise estimator
State-of-charge method of estimation, the flow chart of embodiment is as shown in figure 1, mainly include following steps：
1st, known battery system equivalent-circuit model is determined
1) m × N-shaped battery system
M × N-shaped battery system is to be gone here and there n and to form through m by multiple battery cells, and its structure chart is as shown in Figure 2.For ease of dividing
Analysis, assume in this example that parallel connection type battery system is by 9 battery cells are through 3 strings 3 and form, i.e. 3 × 3 type battery systems.3×3
The rated voltage of each battery cell is 3.7V in type battery system, and rated capacity is 0.86Ah.
2) 3 × 3 type battery system equivalent-circuit models are determined
Battery system equivalent-circuit model (1) is second order equivalent-circuit model, model main circuit by 2 RC parallel circuits,
Controlled voltage source mU0(SOC) and the composition such as internal resistance of cell mR/n, as shown in Figure 3.Battery system performance parameter by with battery list
The relation of body performance parameter is obtained, and as m=3, n=3, specific battery system equivalent-circuit model is as follows：
In above formula, a0~a5Value is respectively -0.915,40.867,3.632,0.537,0.499,0.522, b0~b5Take
Value is respectively 0.1463,30.27,0.1037,0.0584,0.1747,0.1288, c0~c2Value is respectively 0.1063,62.49,
0.0437, d0~d2Value is respectively -200,138,300, e0~e2Value is respectively 0.0712,61.4,0.0288, f0~f2Take
Value is respectively 3083,180,5088.
2nd, cell system space state equation is obtained
1), with battery system SOCbAnd 2 terminal voltage U of RCbs(t)、UblT () is used as system state variables xk, with Ub、Ib
Respectively as system measurements variable ykAnd system input variable, cell system space state equation is set up according to equivalent-circuit model
For
In formula, Ubs、UblIt is 2 RC parallel circuit terminal voltages, τ1、τ2It is time constant, Δ t is sampling
Cycle, ωkIt is system noise, Δ t is the sampling period, and k is the natural number more than 1.
2), according to Kirchhoff's second law, with reference to battery system equivalent-circuit model, battery system output can be obtained and is measured
Equation is：[Ub,k]=3U0,k-RkIb,k-Ubs,k-Ub1,k+υk=gk(xk)+υk=yk, in formula, υkIt is system measurements noise, k is big
In 1 natural number.
3rd, the noise estimation value (4) at k+1 moment is obtained using the intermediateness amount of noise estimator combination last moment, i.e., In formula, Kk+1For gain is joined
Number；ekIt is row difference item, its expression formula isykIt is battery system terminal voltage measured value；dk=(1-b)/(1-bk+1),
B values are 0.995.
4th, by the noise estimation value (4) at k+1 momentFiltered instead of Unscented kalman respectively
Statistical information value (the q of ripple algorithm UKF (5)k+1、Qk+1、rk+1、Rk+1)。
5th, battery system SOC estimations are carried out using Unscented kalman filtering algorithm UKF (5), obtains the middle shape at k+1 moment
State amount (6), i.e.,Pk+1、ek+1。
1) init state variable x averages E () and noise information：x0=E (x0)=[1 0 0], q0=0.1, Q0=0.3,
r0=0.01, R0=1.2) sampled point x is calculatedi,kWith respective weights ω：
In formula, λ=α2(n+h)-n, n=3, α value is that 1, β values are that 2, h values are 0；3) state estimation and mean square error
Time updates：The state estimation time is updated toThe mean square error time is updated toSystem output time is updated to4)
Calculate gain matrix：5) state estimation and mean square error
Measurement updaue：State estimation measurement updaue isMean square error measurement updaue is
Finally, using the intermediateness amount (6) at k+1 moment as subsequent time noise estimator (3) input quantity, meanwhile,
Export the battery charge state estimate SOC at k+1 momentb,k+1, recursion is circulated so as to obtain battery charge shape not in the same time with this
State SOCbEstimate.
System emulation result and Contrast on effect
SOC is carried out to 3 × 3 type battery systems by a kind of this battery system state-of-charge method of estimation with noise estimator
Estimate, while carrying out SOC estimations to this battery system using EKF, contrast to verify sheet one by simulation result and experimental data
Planting the battery system state-of-charge method of estimation with noise estimator has the advantages that fast convergence rate, strong robustness, high precision.
The main constant current operating mode of l-G simulation test, i.e. battery are outwards powered with current constant mode (0.8A).Fig. 4-1~Fig. 4-2 is SOC0When different
Battery constant-current discharge characteristic, wherein Fig. 4-1 is SOC0SOC situations of change when=0.8, Fig. 4-2 is SOC0SOC errors feelings when=0.8
Condition.Knowable to Fig. 4-1, when carrying out SOC estimations using self adaptation UKF (AUKF being designated in figure, similarly hereinafter) and two kinds of algorithms of EKF,
Both of which can preferably match experimental data, but from Fig. 4-2, the error of electric discharge initial time AUKF is smaller, demonstrates this
Method has higher precision；Meanwhile, as shown in Fig. 4-1, at initial time (before 20s), because the noise statisticses information of AUKF is given
Determine initial value differ be set to it is optimal, it is possible to be inaccurate value, thus its estimate SOC value deviate experimental data it is larger, but
After the 20s moment, due to the filter action of noise estimator, its SOC value can tracking test data, two more quickly than EKF algorithm
Person is respectively 20s, 40s at the convergence moment, demonstrates the fast convergence of this method.
Claims (5)
1., the invention discloses a kind of battery system state-of-charge method of estimation with noise estimator, the battery system is by battery
Monomer by m go here and there n and m × N-shaped battery system, wherein m, n be the natural number more than 1,
The described method comprises the following steps：
Cell system space state equation (2) is set up according to known battery system equivalent-circuit model (1) first, noise is recycled
Estimator (3) obtains the noise estimation value (4) at k+1 moment, then with the battery system in cell system space state equation (2)
State-of-charge SOCb, 2 RC parallel circuits terminal voltage as Unscented kalman filtering method UKF (5) state variable, with battery
Input state space equation, the output voltage state space equation of system space state equation (2) are respectively as Unscented kalman
Nonlinear state Equation f () of filter method UKF (5) and measurement equation g (), using the noise estimation value (4) at k+1 moment as
The noise statisticses information of Unscented kalman filtering method UKF (5), the k+1 moment is obtained using Unscented kalman filtering method UKF (5)
Intermediateness amount (6), as the input quantity of subsequent time noise estimator (3), while exporting the battery charge state at k+1 moment
SOCb,k+1, recursion is circulated with this and obtains battery charge state SOCbEstimate.
2. a kind of battery system state-of-charge method of estimation with noise estimator according to claim 1, its feature exists
It is in described noise estimator (3)：
3. a kind of battery system state-of-charge method of estimation with noise estimator according to claim 1, its feature exists
Have in the noise estimation value (4) at described k+1 moment：The state side at k+1 moment is represented respectively
Journey noise Estimation of Mean value, state equation Noise Variance Estimation value, measurement equation noise Estimation of Mean value, measurement equation noise side
Difference estimate.
4. a kind of battery system state-of-charge method of estimation with noise estimator according to claim 1, its feature exists
Have in the intermediateness amount (6) at described k+1 moment：Pk+1、ek+1, represent that the system mode at k+1 moment becomes respectively
Amount estimate, system output variables estimate, noise error covariance, row difference item.
5. a kind of battery system state-of-charge method of estimation with noise estimator according to claim 1, its feature exists
It is applied to m × N-shaped battery system in described method, is also applied for tandem type or parallel connection type battery system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611121601.3A CN106772074A (en) | 2016-12-08 | 2016-12-08 | A kind of battery system state-of-charge method of estimation with noise estimator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611121601.3A CN106772074A (en) | 2016-12-08 | 2016-12-08 | A kind of battery system state-of-charge method of estimation with noise estimator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106772074A true CN106772074A (en) | 2017-05-31 |
Family
ID=58877424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611121601.3A Pending CN106772074A (en) | 2016-12-08 | 2016-12-08 | A kind of battery system state-of-charge method of estimation with noise estimator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106772074A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107831448A (en) * | 2017-11-07 | 2018-03-23 | 国网江苏省电力公司盐城供电公司 | A kind of state-of-charge method of estimation of parallel connection type battery system |
| CN109839596A (en) * | 2019-03-25 | 2019-06-04 | 重庆邮电大学 | SOC estimation method based on the UD adaptive extended kalman filtering decomposed |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105182245A (en) * | 2015-09-08 | 2015-12-23 | 盐城工学院 | High-capacity battery system charge state estimation method based on unscented Kalman filter |
| EP2963434A1 (en) * | 2014-06-30 | 2016-01-06 | Foundation Of Soongsil University-Industry Cooperation | Battery state estimation method and system using dual extended kalman filter, and recording medium for performing the method |
| CN105353315A (en) * | 2015-09-30 | 2016-02-24 | 盐城工学院 | Estimation method of state of charge of battery system on the basis of Unscented Kalman Filter |
| WO2016064198A1 (en) * | 2014-10-24 | 2016-04-28 | 주식회사 엘지화학 | Apparatus for estimating charge state of secondary battery and method therefor |
-
2016
- 2016-12-08 CN CN201611121601.3A patent/CN106772074A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2963434A1 (en) * | 2014-06-30 | 2016-01-06 | Foundation Of Soongsil University-Industry Cooperation | Battery state estimation method and system using dual extended kalman filter, and recording medium for performing the method |
| WO2016064198A1 (en) * | 2014-10-24 | 2016-04-28 | 주식회사 엘지화학 | Apparatus for estimating charge state of secondary battery and method therefor |
| CN105182245A (en) * | 2015-09-08 | 2015-12-23 | 盐城工学院 | High-capacity battery system charge state estimation method based on unscented Kalman filter |
| CN105353315A (en) * | 2015-09-30 | 2016-02-24 | 盐城工学院 | Estimation method of state of charge of battery system on the basis of Unscented Kalman Filter |
Non-Patent Citations (1)
| Title |
|---|
| 黄兵锋等: "基于六参数RC 等效电路模型的锂离子电池SOC 估计", <<重庆交通大学学报( 自然科学版)>> * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107831448A (en) * | 2017-11-07 | 2018-03-23 | 国网江苏省电力公司盐城供电公司 | A kind of state-of-charge method of estimation of parallel connection type battery system |
| CN109839596A (en) * | 2019-03-25 | 2019-06-04 | 重庆邮电大学 | SOC estimation method based on the UD adaptive extended kalman filtering decomposed |
| CN109839596B (en) * | 2019-03-25 | 2021-04-16 | 重庆邮电大学 | SOC estimation method based on UD decomposition and adaptive extended Kalman filtering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105116343B (en) | The electrokinetic cell state of charge method of estimation and system of least square method supporting vector machine | |
| CN104535932B (en) | Lithium ion battery charge state estimating method | |
| CN107390127A (en) | A kind of SOC estimation method | |
| CN104535934B (en) | The electrokinetic cell state of charge method of estimation and system of online feedforward compensation | |
| CN105182245A (en) | High-capacity battery system charge state estimation method based on unscented Kalman filter | |
| CN106405433A (en) | Extended Kalman particle filtering based SOC (State Of Charge) estimation method and system | |
| CN109164392A (en) | A kind of SOC estimation method of power battery | |
| CN105548896A (en) | Power-cell SOC online closed-loop estimation method based on N-2RC model | |
| CN108445421A (en) | A kind of health state of lithium ion battery On-line Estimation method of space-oriented application | |
| CN105353315A (en) | Estimation method of state of charge of battery system on the basis of Unscented Kalman Filter | |
| CN107831448B (en) | A kind of state-of-charge estimation method of parallel connection type battery system | |
| CN107219466A (en) | A kind of lithium battery SOC estimation method for mixing EKF | |
| CN109669132A (en) | A kind of battery charge state estimation method based on variation Bayesian filter | |
| CN113156321B (en) | Estimation method of lithium ion battery state of charge (SOC) | |
| CN106443492A (en) | Method for estimating SOC (State Of Charge) of lithium battery of low-velocity electronic vehicle | |
| CN109839596B (en) | SOC estimation method based on UD decomposition and adaptive extended Kalman filtering | |
| CN107169170B (en) | A kind of prediction technique of battery remaining power | |
| CN105116346A (en) | Series-connected battery system and method for estimating state of charge thereof | |
| CN106772067A (en) | The method that Multiple Time Scales IAPF filters estimated driving force battery charge state and health status | |
| CN107290678A (en) | A kind of electrokinetic cell health status on-line monitoring method | |
| CN105353314A (en) | Estimation method of state of charge of parallel-connected battery system | |
| CN105182246A (en) | Parallel battery system charge state estimation method based on unscented Kalman filter | |
| CN106443496A (en) | Battery charge state estimation method with improved noise estimator | |
| CN112098849A (en) | Lithium battery residual capacity estimation method based on integral Kalman filtering | |
| CN104573294A (en) | Self-adaptive kalman filter estimation algorithm for power battery |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170531 |
|
| WD01 | Invention patent application deemed withdrawn after publication |