CN106815406B - Power battery SOC estimation method based on feature model - Google Patents

Power battery SOC estimation method based on feature model Download PDF

Info

Publication number
CN106815406B
CN106815406B CN201611199513.5A CN201611199513A CN106815406B CN 106815406 B CN106815406 B CN 106815406B CN 201611199513 A CN201611199513 A CN 201611199513A CN 106815406 B CN106815406 B CN 106815406B
Authority
CN
China
Prior art keywords
soc
value
voltage
current
power 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
Application number
CN201611199513.5A
Other languages
Chinese (zh)
Other versions
CN106815406A (en
Inventor
吴珂
卢丹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Automotive Battery Research Institute Co Ltd
Original Assignee
China Automotive Battery Research Institute Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Automotive Battery Research Institute Co Ltd filed Critical China Automotive Battery Research Institute Co Ltd
Priority to CN201611199513.5A priority Critical patent/CN106815406B/en
Publication of CN106815406A publication Critical patent/CN106815406A/en
Application granted granted Critical
Publication of CN106815406B publication Critical patent/CN106815406B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Secondary Cells (AREA)

Abstract

The invention provides a power battery SOC estimation method based on a characteristic model, which comprises the following steps: collecting voltage and/or current values during the operation of the power battery, and determining a resampling coefficient Q according to the collected voltage and/or current values; establishing a characteristic model of the voltage and the SOC of the battery, estimating the SOC value at the next moment according to the current voltage value and the historical SOC value, identifying characteristic coefficients by adopting a recursive least square method with forgetting factors, and comparing the estimated value with an actual value. According to the method, sampling parameters are selected again according to the current and voltage, and when the voltage is at the two ends of the working range, the sampling coefficient Q is 1; when the voltage is in the middle stage of the working range, the two conditions are divided, the calculated amount is reduced when the current is small, and the calculation efficiency is improved; when the current is larger, the sampling quantity is kept unchanged basically, the characteristics of parameters cannot be missed, and the estimation accuracy is ensured.

Description

Power battery SOC estimation method based on feature model
Technical Field
The invention belongs to the field of secondary batteries, and particularly relates to a method for estimating the state of charge of a lithium ion battery.
Background
The State of Charge (SOC) of the battery is also called the remaining capacity, is an important parameter reflecting the battery State, and is also a main basis for the Battery Management System (BMS) of the electric vehicle and the vehicle controller to formulate a control strategy. The SOC is accurately estimated, the battery can be guaranteed to work in a reasonable SOC range, damage to the battery caused by overcharge and overdischarge is prevented, the service life of the battery is prolonged, and the use and maintenance cost is reduced. Besides, the accurate SOC value can enable a user to calculate the driving mileage better and have better driving experience. Therefore, how to accurately and reliably estimate the SOC is an important and difficult task for the battery management system.
In actual estimation, there are two problems that make it more difficult to accurately estimate SOC. On one hand, the SOC cannot be directly measured through a sensor, and the SOC can be estimated only by detecting the voltage, the current, the internal resistance and the temperature of the battery, wherein errors exist in the detection of all parameters; on the other hand, the battery is influenced by factors such as the magnitude of charging and discharging current, temperature, self-discharge, service life and the like during operation, and shows complex nonlinearity, so that the establishment of an accurate battery model is difficult.
In recent years, a large amount of intensive research work has been done by scholars at home and abroad on the estimation of the SOC of the battery, and various estimation methods have been proposed. The open-circuit voltage method, ampere-hour integral method, linear model method, internal resistance method or the combination of the former two methods are proposed earlier; in recent years, some intelligent algorithms, such as kalman filter algorithm, neural network method, fuzzy control algorithm, and the like, are applied to SOC estimation of a battery.
Due to the limitation of hardware conditions, common methods are open circuit voltage method, on-time integration method and the combination algorithm of the two. However, the open-circuit voltage method requires that the battery is left standing for a long enough time, the terminal voltage can be stabilized, and the charging and discharging curves are usually not completely symmetrical and are easy to be confused in the table look-up process, so that the SOC cannot be estimated in real time; the ampere-hour integration method has two problems of current measurement accumulated error and inaccuracy of initial value setting of self-discharge change. The internal resistance method is to determine the SOC by utilizing the functional relationship between the internal resistance of the battery and the electric quantity of the battery, but the internal resistance value of the battery is too small, the accurate measurement is difficult by utilizing a conventional measuring circuit, and a large error exists, so the internal resistance method is not suitable for electric automobiles. The linear model method is to establish a linear equation of the battery voltage, the battery current and the SOC, and obtain the current SOC value by recursion, but is only suitable for the case of low current and slow change of the SOC, and the calculation error of the SOC is also caused by inaccurate initial values. The SOC estimation precision of the neural network method is high, online estimation can be carried out, however, a large amount of data needs to be trained, the existing hardware equipment cannot meet the requirement, the learning time is long, a certain delay is generated, and meanwhile, the precision is greatly influenced by the training method and the training data. The Kalman filtering method is a popular algorithm at present, can overcome accumulated errors of an ampere-hour integral method, and can quickly converge an SOC value to be close to a true value under the condition of large initial errors, but the algorithm has strict requirements on the accuracy of a model, and in practice, the model is time-varying and nonlinear and has large influence on the estimation of the SOC. The fuzzy logic control algorithm has no requirement on the establishment of a model, but needs a great deal of logic experience and is not suitable for the situation of complex vehicle conditions.
Disclosure of Invention
In order to make up for the defects of the algorithm, the invention discloses a power battery SOC estimation method based on a characteristic model, which is used for establishing the characteristic model of a battery, namely, a second-order time-varying differential equation is used for establishing the model of the battery, and a recursive least square method is used for identifying the time-varying coefficient of the characteristic, so that the influence of time variation and nonlinearity on the modeling precision is overcome, the robustness and the adaptability to the abrupt change state are stronger, and the calculated amount is moderate.
The technical scheme for realizing the aim of the invention is as follows:
a power battery SOC estimation method based on a feature model comprises the following steps:
s1, collecting voltage and current values during the operation of the power battery, and determining a resampling coefficient Q according to the collected voltage and current values;
s2, establishing a characteristic model of the voltage and the SOC of the battery, and estimating the SOC value at the next moment according to the voltage value and the SOC value;
s3, identifying the characteristic coefficient by adopting a recursive least square method with a forgetting factor, wherein the initial value of the characteristic coefficient is selected as a zero matrix, and the forgetting factor is a numerical value between 0 and 1;
and S4, comparing the estimated value with the actual value, if the error is larger than the required error, returning to the step S1, changing the Q value, and recalculating.
In the method, the operation of S2 overcomes the problem that the battery is inaccurate due to nonlinearity and time-varying property in the modeling process. S3, by adopting a recursion least square method with forgetting factors, the characteristic coefficient of slow time variation can be identified, and the defect that the traditional least square method is easy to saturate is overcome.
When the power battery is fully charged or emptied, the voltage change is severe, and at the moment, the sampling point is kept unchanged as much as possible; when the voltage value is in the middle stage of the battery voltage range, the judgment is carried out according to the current value, when the current value is smaller, the Q value can be selected to be larger, and when the current value is larger, the Q value is smaller as much as possible. The selection mode accords with the operation characteristics of the power battery, and the calculation amount is properly reduced. The method specifically comprises the following steps:
in step S1, Q is a positive integer greater than or equal to 1, the value of Q is determined according to the current voltage value and current value, and when the power battery is fully charged or emptied, Q is 1; when the voltage value is in the middle stage of the battery voltage range, the judgment is carried out according to the current value, and the two conditions are divided: when the current value is less than 0.5C, Q is selected to be a multiple of 2; when the current value is 0.5C or more, Q is 1 or 2.
Furthermore, when the current voltage value is 1.1-1.2 times greater than the rated voltage value and 0.8-0.7 times less than the rated voltage value, Q is 1; q is selected to be a multiple of 2 and is one of 2, 4, 6 and 8.
In step S2, the SOC value at the next time is estimated from the normalized voltage value and SOC value, which are historical data stored in the battery management system or the charge/discharge device.
In step S2, a feature model is established as follows:
SOC(k+1)＝f1(k)SOC(k)+f2(k)SOC(k-1)+g1(k)U(k) (1)
in the formula (1), f1(k)、f2(k) And g1(k) Is a dynamic slow time varying characteristic coefficient;
SOC (k +1) is the estimated value of SOC, SOC (k) is the current time SOC value, SOC (k-1) is the previous time SOC value, and U (k) is the current time voltage value.
In step S2, the feature model is established as follows:
SOC(k+1)＝f1(k)SOC(k)+f2(k)SOC(k-1)+g1(k)U(k)
＝φT(k)θ(k)
wherein the content of the first and second substances,
φT(k-1)＝[SOC(k) SOC(k-1) U(k)]，θ(k)＝[f1(k) f2(k) g1(k)]T(4)
f1(k)、f2(k) and g1(k) Are dynamic slowly time-varying eigencoefficients.
In step S3, the calculation formula of the recursive least square method with forgetting factor is:
Figure GDA0002385090500000041
wherein f is a forgetting factor, and a numerical value between 0 and 1 is selected; i is a 3-order identity matrix, P is a third-order diagonal matrix,
Figure GDA0002385090500000042
is an estimate of θ (k).
In step S4, the error is calculated by the formula
Figure GDA0002385090500000043
Figure GDA0002385090500000044
And
Figure GDA0002385090500000045
is the estimated value in step S3.
Preferably, in step S4, the required error is set to 2 to 10%. If the error is more than 2-10%, the Q value is increased to carry out recalculation.
The invention has the beneficial effects that:
according to the method, sampling parameters are selected again according to the current and voltage, and when the voltage is at the two ends of the working range, the sampling coefficient Q is 1; when the voltage is in the middle stage of the working range, the two conditions are divided, when the current is small, namely the voltage and the SOC change slowly, the calculation amount can be obviously reduced, and the calculation efficiency is improved; when the current is large, namely the SOC changes rapidly, the sampling number is kept basically unchanged, namely the characteristics of the parameters cannot be missed, and the estimation accuracy can be ensured. Secondly, the characteristic model is a second-order slow time-varying difference equation, the characteristics of the system are compressed in a slow time-varying coefficient, complex non-linearity and high-order characteristics do not need to be considered in the model establishing process, and the modeling efficiency is improved. Finally, the method has moderate calculation amount, improves the estimation precision and can perform online calculation, thereby having good operability and practicability.
Drawings
FIG. 1 is a flow chart of a method for estimating battery SOC based on a feature model according to the present invention;
FIG. 2 is a block diagram of a method for estimating battery SOC based on a feature model;
fig. 3 is a comparison of the curve estimated by the present invention and the curve obtained by the experiment (the rated voltage of the battery pack is 48V).
FIG. 4 illustrates an error curve of the SOC estimation method of the present invention with experimental data.
Detailed Description
The present invention is illustrated by the following preferred embodiments. It will be appreciated by those skilled in the art that the examples are only intended to illustrate the invention and are not intended to limit the scope of the invention.
In the examples, the means used are conventional in the art unless otherwise specified.
Example 1
As shown in fig. 2, the method for estimating the battery SOC based on the feature model of the present invention mainly includes the steps of signal acquisition, resampling coefficient determination, resampling, feature coefficient calculation and feature model estimation, and specifically includes the following steps:
s1, collecting voltage and current values during the operation of the power battery, and determining a resampling coefficient Q according to the collected voltage and/or current; when the sampled voltage value is greater than 1.14 times of the rated voltage value or less than 0.8 time, Q is 1, otherwise, according to the current value, when the current value is less than 0.5C, Q can be selected as a multiple of 2. When the current value is 0.5C or more, the voltage and SOC of the battery change rapidly, and the value of Q is as small as possible, for example, Q is 1 or 2. The formula for determining the Q value is as follows:
Figure GDA0002385090500000061
wherein n is 1,2,3 … ….
And S2, establishing a characteristic model of the voltage and the SOC of the battery, and estimating the SOC value at the next moment according to the current voltage value and the historical SOC value stored in the battery management system.
And S3, identifying the characteristic coefficient by adopting a recursive least square method with a forgetting factor, wherein the initial value of the characteristic coefficient is selected as a zero matrix, and the forgetting factor is a numerical value between 0 and 1.
And S4, comparing the estimated value with the actual value, if the error is larger than the required error, returning to the step S1, changing the Q value, and recalculating.
In the embodiment, the signal acquisition module acquires the voltage value and the current value of the power battery in operation, the resampling coefficient is determined according to the current value and the current voltage value, and when the sampling voltage value is greater than 1.14 times of the rated voltage or less than 0.8 times of the rated voltage, the resampling coefficient is 1; in this embodiment, when the current is within 0.5C, the calculation amount is 1/6, and when the current is greater than 0.5C, the calculation amount is regarded as large current, and Q is 1.
Secondly, the voltage value and the SOC value stored in the battery management system are normalized, and the formula is as follows:
Figure GDA0002385090500000062
setting initial parameters of an identification algorithm, wherein the formula of the identification algorithm is as follows:
Figure GDA0002385090500000071
wherein f is a forgetting factor, and in the embodiment, f is 0.99; i is a 3 rd order identity matrix, P is a third order diagonal matrix, in this embodiment
Figure GDA0002385090500000072
Figure GDA0002385090500000073
Is an estimated value of theta (k),
Figure GDA0002385090500000074
(k represents the time of sampling and the sampling interval is 10ms)
Obtained in the previous step
Figure GDA0002385090500000075
Into the characteristic equation for calculating the SOC, as follows:
Figure GDA0002385090500000076
in the formula, SOCEstimating(k +1) is an estimated value of SOC obtained by estimating based on the SOC values of the current time SOC (k) and the previous time SOC (k-1) and the voltage value U (k) of the current time (i.e., "new value" calculated by equation (5)),
Figure GDA0002385090500000077
and
Figure GDA0002385090500000078
is the coefficient estimated in the previous step.
Fifthly, the parameters identified in the previous step
Figure GDA0002385090500000079
Convergence, and error calculation are performed, the calculation formula is as follows:
Figure GDA00023850905000000710
in the formula, e (k) is an error value between the actual SOC and the estimated SOC, which is used to verify whether the method meets the actual requirement, and the error requirement is set to 5% in this embodiment.
The estimation result and error of the charging process are shown in fig. 3 and 4, respectively, and it can be seen from fig. 3 and 4 that the estimation error of the system has converged to within 3% when the SOC is 0.1, i.e. 10%, and a resampling process is added before the calculation, so that the amount of calculation is reduced and the calculation speed is increased. In the actual work of the battery pack, the SOC range generally used is 20% to 80%, so that the SOC estimation is fast and high in accuracy, and the requirement in actual use is met.
The above examples are only for describing the preferred embodiments of the present invention, and are not intended to limit the scope of the present invention, and various modifications and improvements made to the technical solution of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims (7)

1. A power battery SOC estimation method based on a feature model is characterized by comprising the following steps:
s1, collecting voltage and current values during the operation of the power battery, and determining a resampling coefficient Q according to the collected voltage and current values;
s2, establishing a characteristic model of the voltage and the SOC of the battery according to the SOC value at the current moment, the SOC value at the previous moment and the voltage value at the current moment, and estimating the SOC value at the next moment according to the voltage value and the SOC value;
s3, identifying the characteristic coefficient by adopting a recursive least square method with a forgetting factor, wherein the initial value of the characteristic coefficient is selected as a zero matrix, and the forgetting factor is a numerical value between 0 and 1;
s4, comparing the estimated value with the actual value, if the error is larger than the required error, returning to the step S1, changing the Q value, and recalculating;
in step S2, a feature model is established as follows:
SOC(k+1)＝f1(k)SOC(k)+f2(k)SOC(k-1)+g1(k)U(k)
in the formula (1), f1(k)、f2(k) And g1(k) Is a dynamic slow time varying characteristic coefficient;
SOC (k +1) is an estimated value of SOC, SOC (k) is an SOC value at the current moment, SOC (k-1) is an SOC value at the previous moment, and U (k) is a voltage value at the current moment; alternatively, the first and second electrodes may be,
in step S2, the feature model is established as follows:
SOC(k+1)＝f1(k)SOC(k)+f2(k)SOC(k-1)+g1(k)U(k)
＝φT(k)θ(k)
wherein phi isT(k)＝[SOC(k) SOC(k-1) U(k)]，θ(k)＝[f1(k) f2(k) g1(k)]T，f1(k)、f2(k) And g1(k) Are dynamic slowly time-varying eigencoefficients.
2. The power battery SOC estimation method according to claim 1, wherein in step S1, Q is a positive integer greater than or equal to 1, the value of Q is determined according to the current voltage value and current value, and Q is 1 when the power battery is fully charged or emptied; when the voltage value is in the middle stage of the battery voltage range, the judgment is carried out according to the current value, and the two conditions are divided: when the current value is less than 0.5C, Q is selected to be a multiple of 2; when the current value is 0.5C or more, Q is 1 or 2.
3. The SOC estimation method for the power battery according to claim 2, wherein when the current voltage value is 1.1-1.2 times greater than the rated voltage value and 0.8-0.7 times less than the rated voltage value, Q is 1; when Q is selected as a multiple of 2, Q is one of 2, 4, 6 and 8.
4. The power battery SOC estimation method according to claim 1, wherein in step S2, the SOC value at the next time is estimated from the voltage value and the SOC value subjected to the normalization processing, the voltage value and the SOC value used for the normalization processing being history data saved in the battery management system or the charge and discharge device.
5. The method for estimating SOC of a power battery according to claim 1, wherein in step S3, the calculation formula of recursive least squares with forgetting factor is:
Figure FDA0002536640390000021
wherein f is a forgetting factor, and a numerical value between 0 and 1 is selected; i is a 3-order identity matrix, P is a third-order diagonal matrix,
Figure FDA0002536640390000022
is an estimate of θ (k).
6. The method according to claim 5, wherein in step S4, the error is calculated by the formula:
Figure FDA0002536640390000023
wherein the content of the first and second substances,
Figure FDA0002536640390000024
and
Figure FDA0002536640390000025
is the estimated value in step S3.
7. The method for estimating SOC of a power battery according to any one of claims 1-6, wherein in step S4, the required error is set to 2-10%.
CN201611199513.5A 2016-12-22 2016-12-22 Power battery SOC estimation method based on feature model Active CN106815406B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611199513.5A CN106815406B (en) 2016-12-22 2016-12-22 Power battery SOC estimation method based on feature model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611199513.5A CN106815406B (en) 2016-12-22 2016-12-22 Power battery SOC estimation method based on feature model

Publications (2)

Publication Number Publication Date
CN106815406A CN106815406A (en) 2017-06-09
CN106815406B true CN106815406B (en) 2020-08-04

Family

ID=59109508

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611199513.5A Active CN106815406B (en) 2016-12-22 2016-12-22 Power battery SOC estimation method based on feature model

Country Status (1)

Country Link
CN (1) CN106815406B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111929585B (en) * 2019-05-13 2023-08-04 丰翼科技(深圳)有限公司 Battery charge state calculating device, method, server and medium
CN116338460B (en) * 2023-04-11 2024-04-09 宁波禾旭汽车科技有限公司 New energy automobile battery allowance identification system based on multi-parameter analysis

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104573401A (en) * 2015-02-03 2015-04-29 中国科学院宁波材料技术与工程研究所 Method and device for estimating state of charge of battery

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10222426B2 (en) * 2013-06-14 2019-03-05 Hrl Laboratories, Llc Double-pulse technique for on-line diagnostics of electrochemical systems

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104573401A (en) * 2015-02-03 2015-04-29 中国科学院宁波材料技术与工程研究所 Method and device for estimating state of charge of battery

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A real-time joint estimator for model parameters and state of charge of lithium-ion batteries in electric vehicles;Gao J 等;《Energies》;20151231;第8卷(第8期);第8594-8612页 *
锂电池化成参数的检测与估计技术研究;李明飞;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20160315;第C042-1762页 *

Also Published As

Publication number Publication date
CN106815406A (en) 2017-06-09

Similar Documents

Publication Publication Date Title
CN106716158B (en) Battery charge state evaluation method and device
CN107991623B (en) Battery ampere-hour integral SOC estimation method considering temperature and aging degree
CN107402353B (en) Method and system for carrying out filtering estimation on state of charge of lithium ion battery
CN108717164B (en) SOC calibration method and system for battery
CN107843852B (en) A method of off-line data is obtained to estimate battery united state
WO2021181536A1 (en) Deterioration degree diagnosis device
WO2015106691A1 (en) Soc estimation method for power battery for hybrid electric vehicle
CN108732508B (en) Real-time estimation method for lithium ion battery capacity
CN105425153B (en) A kind of method of the state-of-charge for the electrokinetic cell for estimating electric vehicle
CN108107372A (en) Accumulator health status quantization method and system based on the estimation of SOC subregions
CN112838631A (en) Dynamic charging management and control device for power battery and charging diagnosis method for power battery
CN104535932A (en) Lithium ion battery charge state estimating method
CN105021996A (en) Battery SOH (section of health) estimation method of energy storage power station BMS (battery management system)
CN109358293B (en) Lithium ion battery SOC estimation method based on IPF
CN110795851A (en) Lithium ion battery modeling method considering environmental temperature influence
CN111337832A (en) Power battery multidimensional fusion SOC and SOH online joint estimation method
CN105044606B (en) A kind of SOC methods of estimation based on parameter adaptive battery model
CN108445422B (en) Battery state of charge estimation method based on polarization voltage recovery characteristics
CN111856282B (en) Vehicle-mounted lithium battery state estimation method based on improved genetic unscented Kalman filtering
CN109901072B (en) Retired battery parameter detection method based on historical data and laboratory test data
CN111781503A (en) Lithium ion energy storage battery SOC online estimation method
CN107102268A (en) A kind of battery rate of charge evaluation method of battery management system
CN115389936A (en) Online prediction method for continuous peak power capability of digital-analog hybrid driven lithium battery
CN111216595B (en) SOC calibration method of severe hybrid electric vehicle based on lithium battery equivalent circuit model
CN110988709A (en) SOE and SOP joint estimation method for battery management system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CB03 Change of inventor or designer information
CB03 Change of inventor or designer information

Inventor after: Li Ang

Inventor after: Tang Haibo

Inventor after: Wu Ke

Inventor after: Lu Dan

Inventor before: Wu Ke

Inventor before: Lu Dan