CN115902667A - Lithium battery SOC estimation method based on weight and volume point self-adaption - Google Patents

Lithium battery SOC estimation method based on weight and volume point self-adaption Download PDF

Info

Publication number
CN115902667A
CN115902667A CN202310114464.4A CN202310114464A CN115902667A CN 115902667 A CN115902667 A CN 115902667A CN 202310114464 A CN202310114464 A CN 202310114464A CN 115902667 A CN115902667 A CN 115902667A
Authority
CN
China
Prior art keywords
state
battery
value
weight
moment
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202310114464.4A
Other languages
Chinese (zh)
Other versions
CN115902667B (en
Inventor
宁雪峰
芦大伟
李龙
韦薇
姚俊钦
袁炜灯
王永源
李元佳
刘贯科
张海鹏
陈鹏
陈文睿
秦立斌
钟荣富
林志强
蒋紫薇
戴喜良
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.)
Dongguan Power Supply Bureau of Guangdong Power Grid Co Ltd
Original Assignee
Dongguan Power Supply Bureau of Guangdong Power Grid 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 Dongguan Power Supply Bureau of Guangdong Power Grid Co Ltd filed Critical Dongguan Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority to CN202310114464.4A priority Critical patent/CN115902667B/en
Publication of CN115902667A publication Critical patent/CN115902667A/en
Application granted granted Critical
Publication of CN115902667B publication Critical patent/CN115902667B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/10—Energy storage using batteries

Landscapes

  • Tests Of Electric Status Of Batteries (AREA)

Abstract

The invention discloses a lithium battery SOC estimation method based on weight and volume point self-adaption. Firstly, acquiring voltage and current data of a lithium battery under different working conditions through voltage and current sensors, and establishing a second-order battery equivalent circuit model; establishing a state equation of a battery model, and identifying resistance-capacitance parameters of an equivalent circuit model; and establishing a nonlinear equation of a discrete time state and a measurement state of the second-order model, and estimating the SOC of the lithium battery by adopting a Kalman filtering algorithm with adaptive weight and volume points. The method eliminates linearization errors, reduces calculation time, simultaneously avoids the condition that SOC estimation is not accurate enough when the problems of battery model errors, unknown measurement noise characteristics and the like exist, greatly improves robustness, solves the problems that the capacity points and the weight of the traditional capacity Kalman filtering algorithm are fixed and unchangeable, has the advantages of high precision and strong robustness, and is suitable for SOC estimation in the battery management system of the energy storage power station.

Description

Lithium battery SOC estimation method based on weight and volume point self-adaption
Technical Field
The invention belongs to the technical field of battery management of energy storage power stations, and relates to a Kalman filtering lithium battery SOC estimation method based on weight and volume point self-adaptation.
Background
Lithium ion batteries are widely used in energy storage power stations due to their advantages of high energy, long service life, low self-discharge rate, etc. The State of Charge (SOC) estimation of a battery is one of key technologies of a battery management system of an energy storage power station, and plays an important role in protection of the battery, prediction of service life, thermal management and the like.
At present, common SOC estimation methods comprise an open-circuit voltage method and a resistance method based on experiments, but the methods need longer test time and are not beneficial to practical application. The data-driven methods include a support vector machine, a convolutional neural network and a long-short-term neural network, and the methods need a large number of experimental samples and are large in calculation processing. The method based on the battery model comprises an equivalent circuit model and an electrochemical model, the model can better reflect the dynamic characteristics and the static characteristics of the battery, and an equation established by the model has the characteristic of nonlinearity, so that the method for efficiently processing the nonlinear equation is very important. The cubature Kalman filtering algorithm is commonly used in application scenes such as smooth filtering, integrated navigation, trajectory tracking and the like at present, and has a good effect on processing a nonlinear equation. The battery is a highly nonlinear system, and the volumetric Kalman filtering is very suitable for estimation of the SOC. However, the conventional kalman filter has the following two problems: (1) When the dimension is larger, the distance between the volume point and the central point is increased, so that the filter is easy to have a non-local sampling problem; (2) The weights of all volume points are the same, which is easy to cause estimation error. Therefore, the prediction and error covariance obtained from each volume point cannot accurately reflect the actual statistical characteristics of the system error, thereby affecting the estimation accuracy of the filter.
The invention provides a Kalman filtering lithium battery SOC estimation method based on weight and volume point self-adaptation. The method obviously improves the estimation precision under the condition of not increasing a large amount of calculation. The method comprises the steps of firstly establishing a second-order equivalent circuit model of the lithium battery, identifying resistance-capacitance parameters by using a recursive least square method with forgetting factors, and then estimating the state of charge of the lithium battery by using a weight and volume point adaptive Kalman filtering method. And finally, obtaining an accurate SOC estimation value through continuous iteration.
Disclosure of Invention
The invention aims to provide a Kalman filtering lithium battery SOC estimation method based on weight and volume point self-adaptation, and provides a technical scheme for accurate estimation of the state of charge of a lithium ion battery. Firstly, a second-order equivalent circuit model of the lithium battery is established, resistance-capacitance parameters are identified by using a recursive least square method with forgetting factors, and then the state of charge of the lithium battery is estimated by using a Kalman filtering method with adaptive weight and volume points. And finally, obtaining an accurate SOC estimation value through continuous iteration, wherein the method can reduce errors caused by estimation and improve the estimation precision and robustness.
The invention is realized by the following technical scheme:
1. acquiring current and voltage test data and an initial battery SOC value of the lithium battery under different working conditions, and establishing an equivalent circuit model of the lithium battery;
the test voltage and current data of the battery are mainly obtained through a battery test system, the initial value of SOC is provided by a battery manufacturer, and the main working conditions of the SOC comprise constant current charge and discharge test, mixed power pulse (HPPC) working condition test and dynamic stress working condition test (DST). Because the second-order RC equivalent circuit model can better simulate the dynamic and static characteristics of the battery, the second-order RC equivalent circuit is adopted as the battery model of the energy storage power station.
2. Establishing a state equation of a battery model;
obtaining a state equation of the lithium battery model according to kirchhoff voltage and current law by adopting a second-order RC equivalent circuit model;
Figure SMS_1
（1）
in the formula (1), the reaction mixture is,
Figure SMS_3
is the open circuit voltage of the battery>
Figure SMS_4
Based on the battery terminal voltage>
Figure SMS_5
Is the ohmic internal resistance of the lithium battery,/>
Figure SMS_6
is the operating current of the battery>
Figure SMS_8
And &>
Figure SMS_9
Is a polarization resistance, is->
Figure SMS_11
And &>
Figure SMS_7
Is a polarized capacitor>
Figure SMS_10
And &>
Figure SMS_12
Is a polarization resistance>
Figure SMS_13
And &>
Figure SMS_14
The corresponding voltage->
Figure SMS_15
And &>
Figure SMS_16
Are respectively based on>
Figure SMS_17
And &>
Figure SMS_2
The derivative of (c).
3. Performing equivalent circuit model resistance-capacitance parameters
Figure SMS_18
Identification, in which>
Figure SMS_19
Is ohmic internal resistance, and is greater or less than>
Figure SMS_20
And &>
Figure SMS_21
Is a polarization resistance, is->
Figure SMS_22
And &>
Figure SMS_23
Is a polarization capacitance.
Method for performing equivalent circuit model resistance-capacitance parameters by using least square method with forgetting factor recursion
Figure SMS_24
And identifying the obtained corresponding resistance-capacitance value.
4. Establishing a discrete time state and measurement state nonlinear equation of a second-order model;
and establishing a discrete time state and measurement state nonlinear equation of a second-order model. Selecting according to the ampere-hour integral equation (2) and the battery state equation (1)
Figure SMS_25
As the state variable, the lithium ion battery state equation (3) and the measurement equation (4) can be listed through discretization.
Figure SMS_26
(2)
Figure SMS_27
(3)
Figure SMS_28
(4)
In the formulae (2), (3) and (4),
Figure SMS_32
and &>
Figure SMS_34
SOC values of the battery at the time k and the time k-1, respectively>
Figure SMS_36
For the maximum available capacity of the battery>
Figure SMS_37
For coulombic efficiency, is>
Figure SMS_40
Is a sampling period, is>
Figure SMS_41
Is a time constant->
Figure SMS_43
And &>
Figure SMS_29
Polarization resistance->
Figure SMS_33
The corresponding voltage->
Figure SMS_35
And &>
Figure SMS_38
Polarization resistance->
Figure SMS_39
The corresponding voltage->
Figure SMS_42
An open circuit voltage corresponding to the SOC value of the battery at the time k>
Figure SMS_44
Terminal voltage of the battery at time k->
Figure SMS_45
For the operating current of the battery at the moment k>
Figure SMS_30
Is process noise->
Figure SMS_31
To observe the noise. />
(a) Value of initialized state variable
Figure SMS_46
Process noise covariance (— er)>
Figure SMS_47
) Measuring the noise covariance (4 @)>
Figure SMS_48
) And state error covariance (< >>
Figure SMS_49
)；
(b) Calculating a mean of values of state variables
Figure SMS_50
And combining the state error covariance (< >>
Figure SMS_51
) Singular value decomposition is carried out, and the cosine similarity (is calculated>
Figure SMS_52
) The decomposition and calculation method is as follows:
Figure SMS_53
(5)
in the formula
Figure SMS_56
Is state error covariance ^ h->
Figure SMS_59
Column matrix->
Figure SMS_61
Based on the expectation of the value of the state variable>
Figure SMS_63
Is the mean value of the state variable>
Figure SMS_65
In order to be the covariance of the state error, device for combining or screening>
Figure SMS_67
Is->
Figure SMS_69
Three matrices obtained by singular value decomposition->
Figure SMS_54
Is/>
Figure SMS_57
Is greater than or equal to>
Figure SMS_58
Is->
Figure SMS_60
Is determined by the feature vector of (a), device for selecting or keeping>
Figure SMS_62
Is->
Figure SMS_64
Transposed of (5)>
Figure SMS_66
Is->
Figure SMS_68
Is transposed matrix of->
Figure SMS_55
Is a diagonal matrix;
(c) Determining a height volume criterion by using a high-order radial criterion, and generating corresponding volume points according to the cosine similarity and the height volume criterion
Figure SMS_70
And weight +>
Figure SMS_71
：
Figure SMS_72
(6)
In the formula
Figure SMS_73
,/>
Figure SMS_74
Is->
Figure SMS_75
Is greater than or equal to>
Figure SMS_76
A column matrix; />
Figure SMS_77
Is the dimension of the state equation, <' > is>
Figure SMS_78
Is a variable constant, typically taken to be 1.6.
(d) Calculating a state variable predicted value and a state error covariance value;
Figure SMS_79
(7)
in the formula
Figure SMS_95
Is the state equation, <' >>
Figure SMS_96
Is the first->
Figure SMS_97
At a moment in time>
Figure SMS_98
Is->
Figure SMS_99
A status variable predictor value at the time instant>
Figure SMS_100
For an error covariance predictor, <' >>
Figure SMS_101
Is->
Figure SMS_80
At a moment in time +>
Figure SMS_82
The weight of a respective volume point->
Figure SMS_84
Is->
Figure SMS_86
Is at a moment->
Figure SMS_88
Each volume point is selected and/or judged>
Figure SMS_89
Is->
Figure SMS_92
Transposition of the status variable predictor at a time instant>
Figure SMS_94
Is->
Figure SMS_81
Time course noise covariance matrix,/>>
Figure SMS_83
Is->
Figure SMS_85
At a moment in time +>
Figure SMS_87
Transposition of the state function values of individual volume points,. According to the value of the volume point, the value is greater than or equal to>
Figure SMS_90
Is->
Figure SMS_91
Is at a moment->
Figure SMS_93
The value of the state function for each volume point.
(e) And (d) performing volume point calculation and weight calculation again by using the formula (6) according to the state variable predicted value and the state error covariance predicted value obtained in the step (d).
(f) Updating the measurement predicted value and the measurement autocorrelation and cross-correlation covariance value;
Figure SMS_102
(8)
in the formula
Figure SMS_119
For the measurement equation, <' >>
Figure SMS_120
Is->
Figure SMS_121
The measured predicted value at the moment is greater or less than>
Figure SMS_122
Is->
Figure SMS_123
Is at a moment->
Figure SMS_124
The weight of a respective volume point->
Figure SMS_125
Is->
Figure SMS_103
The moment measured autocorrelation error covariance matrix, <' > is then evaluated>
Figure SMS_105
Is->
Figure SMS_107
Moment measurement cross-correlation error covariance matrix, <' >>
Figure SMS_109
Is->
Figure SMS_111
At a moment in time +>
Figure SMS_113
Each volume point is selected and/or judged>
Figure SMS_115
Is->
Figure SMS_118
The transfer of the measured predicted value of the time>
Figure SMS_104
Is->
Figure SMS_106
The measured noise covariance matrix at a time @>
Figure SMS_108
Is->
Figure SMS_110
Is at a moment->
Figure SMS_112
Transposition of the measurement function values of individual volume points,. According to the value of the volume point, then the value is greater or less than>
Figure SMS_114
Is->
Figure SMS_116
Is at a moment->
Figure SMS_117
The measurement function value of each volume point.
(g) Calculating gain
Figure SMS_126
Updating state estimate and state covariance estimateA value;
Figure SMS_127
；
in the formula
Figure SMS_129
Is->
Figure SMS_131
The voltage data measured by the battery management system is based on the time>
Figure SMS_132
Is->
Figure SMS_134
The time state error covariance matrix, < > >>
Figure SMS_136
Is->
Figure SMS_137
Time-of-day measured autocorrelation error covariance matrix,/>>
Figure SMS_139
Is the gain matrix at time k +1>
Figure SMS_128
Transposing the gain matrix for the time k +1, superscript @>
Figure SMS_130
Stands for transposed, subscript->
Figure SMS_133
Represents a fifth->
Figure SMS_135
At that moment, is greater or less>
Figure SMS_138
Is the value of the state variable at the moment k +1>
Figure SMS_140
Is->
Figure SMS_141
Moment measurement cross-correlation error covariance matrix, <' >>
Figure SMS_142
Error covariance prediction.
(h) And (c) repeating the processes from (b) to (g), calculating the state variable and the state covariance at the next moment until the operation is finished, and extracting the result to obtain the SOC estimated value.
The invention has the following beneficial effects:
the lithium battery SOC estimation method based on the weight and volume point self-adaption provides a technical scheme for accurate estimation of the lithium battery SOC. The adaptability of the volume points and the weight in the fine cubature Kalman filtering is realized, and the volume points and the weight are continuously updated in each iteration process, so that the method has important practical significance for improving the SOC estimation precision and reducing the calculation time.
Drawings
Fig. 1 is a flow chart of a weight and volume point adaptive kalman filter lithium battery SOC estimation method.
Fig. 2 is a second-order equivalent circuit diagram of the lithium battery.
Detailed description of the preferred embodiments
The present invention will be described in detail below with reference to the drawings and examples.
Referring to fig. 1, an embodiment of the present invention provides a lithium battery SOC estimation method based on weight and volume point self-adaptation, including the following steps:
1. obtaining current and voltage test data and an initial SOC value of the lithium battery under different working conditions, and establishing an equivalent circuit model of the lithium battery.
The test voltage and current data of the battery are mainly obtained through a battery test system, the initial value of SOC is provided by a battery manufacturer, and the main working conditions of the SOC comprise constant current charge and discharge test, mixed power pulse (HPPC) working condition test and dynamic stress working condition test (DST). Because the second-order RC equivalent circuit model can better simulate the dynamic and static characteristics of the battery, the second-order RC equivalent circuit is adopted as the battery model of the energy storage power station.
2. And establishing a state equation of the battery model.
Referring to fig. 2, a second-order RC equivalent circuit model is adopted to obtain a state equation of a lithium battery model according to kirchhoff voltage and current law;
Figure SMS_143
（1）
in the formula (1), the acid-base catalyst,
Figure SMS_146
is the open circuit voltage of the battery>
Figure SMS_148
Based on the battery terminal voltage>
Figure SMS_150
Ohmic internal resistance of lithium battery, and based on the measured value>
Figure SMS_152
Is the working current of the battery>
Figure SMS_154
And &>
Figure SMS_156
Is a polarization resistance->
Figure SMS_158
And &>
Figure SMS_144
Is a polarized capacitor>
Figure SMS_147
And &>
Figure SMS_149
Is a polarization resistance->
Figure SMS_151
And &>
Figure SMS_153
The corresponding voltage->
Figure SMS_155
And &>
Figure SMS_157
Are respectively based on>
Figure SMS_159
And &>
Figure SMS_145
The derivative of (c).
3. Performing equivalent circuit model resistance-capacitance parameters
Figure SMS_160
Identification, in which>
Figure SMS_161
Is ohmic internal resistance, and is greater or less than>
Figure SMS_162
And &>
Figure SMS_163
Is a polarization resistance, is->
Figure SMS_164
And &>
Figure SMS_165
Is a polarization capacitance.
Method for performing equivalent circuit model resistance-capacitance parameters by using least square method with forgetting factor recursion
Figure SMS_166
And identifying the obtained corresponding resistance-capacitance value.
4. Establishing a discrete time state and measurement state nonlinear equation of a second-order model;
and establishing a discrete time state and measurement state nonlinear equation of a second-order model. Selecting according to an ampere-hour integral equation (2) and a battery state equation (1)
Figure SMS_167
As the state variable, the lithium ion battery state equation (3) and the measurement equation (4) can be listed through discretization.
Figure SMS_168
(2)
Figure SMS_169
(3)
Figure SMS_170
(4)
In the formulae (2), (3) and (4),
Figure SMS_176
and &>
Figure SMS_178
SOC values of the battery at the time k and the time k-1, respectively>
Figure SMS_181
For the maximum available capacity of the battery, is selected>
Figure SMS_183
For coulomb efficiency, <' > based on>
Figure SMS_185
For a sampling period, <' >>
Figure SMS_186
Is a time constant->
Figure SMS_187
And &>
Figure SMS_171
Polarization resistance->
Figure SMS_173
Corresponding voltage, < '> or <' > is combined>
Figure SMS_174
And &>
Figure SMS_177
Polarization resistance for time k and for time k-1, respectively>
Figure SMS_179
The corresponding voltage->
Figure SMS_180
Is the open-circuit voltage corresponding to the SOC value of the battery at the moment k>
Figure SMS_182
Terminal voltage of the battery at time k->
Figure SMS_184
For the operating current of the battery at the moment k>
Figure SMS_172
Is a process noise, is asserted>
Figure SMS_175
To observe the noise.
5. Estimating the SOC of the lithium battery by adopting a Kalman filtering algorithm with adaptive weight and volume points;
(a) Value of initialized state variable
Figure SMS_188
Process noise covariance (— er)>
Figure SMS_189
) Measuring the noise covariance (4 @)>
Figure SMS_190
) And state error covariance (< >>
Figure SMS_191
)；
(b) Calculating a mean of values of state variables
Figure SMS_192
And combining the state error covariance (< >>
Figure SMS_193
) Performing singular value decomposition and calculating cosine similarity (< >>
Figure SMS_194
). The decomposition and calculation method is as follows:
Figure SMS_195
(5)
in the formula
Figure SMS_200
Is the state error covariance>
Figure SMS_201
Is/are>
Figure SMS_204
Column matrix->
Figure SMS_206
Based on the expectation of the value of the state variable>
Figure SMS_208
Is the mean value of the state variable>
Figure SMS_210
Is state error covariance->
Figure SMS_212
Is->
Figure SMS_196
Three matrices obtained by singular value decomposition->
Figure SMS_198
Is/>
Figure SMS_202
Is greater than or equal to>
Figure SMS_203
Is->
Figure SMS_205
Is greater than or equal to>
Figure SMS_207
Is->
Figure SMS_209
Transposed of (5)>
Figure SMS_211
Is->
Figure SMS_197
Is transposed matrix of->
Figure SMS_199
Is a diagonal matrix.
(c) Determining a new height volume rule by using a high-order radial rule, and generating corresponding volume points according to cosine similarity and a new volume rule
Figure SMS_213
And the weight->
Figure SMS_214
：
Figure SMS_215
(6)
In the formula
Figure SMS_217
,/>
Figure SMS_219
Is->
Figure SMS_221
Or a number of>
Figure SMS_223
A column matrix; />
Figure SMS_225
Dimension of the equation of state>
Figure SMS_227
Is a variable constant, generally taken to be 1.6, ° v>
Figure SMS_228
Is a first->
Figure SMS_230
Each volume point is selected and/or judged>
Figure SMS_233
Is the first->
Figure SMS_235
Each volume point is selected and/or judged>
Figure SMS_236
Is the first->
Figure SMS_237
Each volume point is selected and/or judged>
Figure SMS_238
Is the first->
Figure SMS_239
Each volume point is selected and/or judged>
Figure SMS_240
Is a first->
Figure SMS_216
The weight of a respective volume point->
Figure SMS_218
Is the first->
Figure SMS_220
A volume point weight, based on the weight of the volume point>
Figure SMS_222
Is the first->
Figure SMS_224
A volume point weight, based on the weight of the volume point>
Figure SMS_226
Is the first->
Figure SMS_229
Individual volume point weights. />
Figure SMS_231
Is->
Figure SMS_232
Is also based on the probability value of>
Figure SMS_234
And (4) weighting values.
(d) Calculating a state variable predicted value and a state error covariance value;
Figure SMS_241
(7)
in the formula
Figure SMS_257
Is the state equation, <' >>
Figure SMS_258
Is a first->
Figure SMS_259
At that moment, is greater or less>
Figure SMS_260
Is->
Figure SMS_261
A status variable predictor value at the time instant>
Figure SMS_262
For an error covariance predictor, <' >>
Figure SMS_263
Is->
Figure SMS_242
Is at a moment->
Figure SMS_245
The weight of a respective volume point->
Figure SMS_247
Is->
Figure SMS_249
Is at a moment->
Figure SMS_250
Each volume point is selected and/or judged>
Figure SMS_253
Is->
Figure SMS_254
Transposition of the status variable predictor at a time instant>
Figure SMS_256
Is->
Figure SMS_243
The time of day process noise covariance matrix, <' >>
Figure SMS_244
Is->
Figure SMS_246
At a moment in time +>
Figure SMS_248
Transposition of the state function values of individual volume points,. According to the value of the volume point, the value is greater than or equal to>
Figure SMS_251
Is->
Figure SMS_252
At a moment in time +>
Figure SMS_255
The value of the state function for each volume point.
(e) Performing volume point calculation and weight calculation again by using an equation (6) according to the state variable predicted value and the state error covariance predicted value obtained in the step (d);
(f) Updating the measurement predicted value and the measurement autocorrelation and cross-correlation covariance value;
Figure SMS_264
(8)
in the formula
Figure SMS_281
For the measurement equation, <' >>
Figure SMS_282
Is->
Figure SMS_283
The measured predicted value at the moment is greater or less than>
Figure SMS_284
Is->
Figure SMS_285
Is at a moment->
Figure SMS_286
The weight of a respective volume point->
Figure SMS_287
Is->
Figure SMS_265
The moment measured autocorrelation error covariance matrix, <' > is then evaluated>
Figure SMS_267
Is->
Figure SMS_269
Moment measurement cross-correlation error covariance matrix, <' >>
Figure SMS_271
Is->
Figure SMS_273
Is at a moment->
Figure SMS_275
Each volume point is selected and/or judged>
Figure SMS_278
Is->
Figure SMS_280
The transfer of the measured predicted value of the time>
Figure SMS_266
Is composed of
Figure SMS_268
The measured noise covariance matrix at a time @>
Figure SMS_270
Is->
Figure SMS_272
Is at a moment->
Figure SMS_274
The transpose of the measurement function values of each volume point,
Figure SMS_276
is->
Figure SMS_277
Is at a moment->
Figure SMS_279
The measurement function value of each volume point.
(g) Calculating gain
Figure SMS_288
Updating the state estimation value and the state covariance estimation value;
Figure SMS_289
；
in the formula
Figure SMS_291
Is->
Figure SMS_294
The voltage data measured by the battery management system is based on the time>
Figure SMS_296
Is->
Figure SMS_298
The time state error covariance matrix, < > >>
Figure SMS_299
Is->
Figure SMS_301
The moment measured autocorrelation error covariance matrix, <' > is then evaluated>
Figure SMS_302
Is the gain matrix at time k +1>
Figure SMS_290
Transposing the gain matrix for the moment k +1, superscript @>
Figure SMS_292
Stands for transposed, subscript->
Figure SMS_293
Represents a fifth->
Figure SMS_295
At that moment, is greater or less>
Figure SMS_297
Is the value of the state variable at the moment k +1>
Figure SMS_300
Is->
Figure SMS_303
Moment measurement cross-correlation error covariance matrix, <' >>
Figure SMS_304
Error covariance prediction.
(h) And (c) repeating the processes from (b) to (g), calculating the state variable and the state covariance at the next moment until the operation is finished, and extracting the result to obtain the SOC estimated value.
The above is only a description of the preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above examples, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. Various modifications, additions and substitutions for the specific embodiments described herein may be made by those skilled in the art without departing from the spirit and principles of the invention.

Claims (6)

1. A lithium battery SOC estimation method based on weight and volume point self-adaption comprises the following steps:
A. acquiring current and voltage test data and an initial battery SOC value of the lithium battery under different working conditions, and establishing an equivalent circuit model of the lithium battery;
B. establishing a state equation of a battery model;
C. performing equivalent circuit model resistance-capacitance parameters
Figure QLYQS_1
Identification, in which>
Figure QLYQS_2
Ohmic internal resistance, based on the measured value>
Figure QLYQS_3
And &>
Figure QLYQS_4
Is a polarization resistance, is->
Figure QLYQS_5
And &>
Figure QLYQS_6
Is a polarization capacitor;
D. establishing a discrete time state and measurement state nonlinear equation of a second-order model;
E. and estimating the SOC of the lithium battery by adopting a Kalman filtering algorithm with adaptive weight and volume points.
2. The lithium battery SOC estimation method based on weight and volume point self-adaptation according to claim 1, wherein in the step A, the test voltage and current data of the battery are obtained through a battery test system, the initial value of SOC is provided by a battery manufacturer, and the working condition test items comprise a constant current charge and discharge test, a mixed power pulse working condition test HPPC and a dynamic stress working condition test DST.
3. The weight and volume point self-adaptive lithium battery SOC estimation method according to claim 1, wherein in the step B, a second-order RC equivalent circuit model is adopted, and a state equation of a lithium battery model is obtained according to kirchhoff's voltage and current law;
Figure QLYQS_7
（1）
in the formula (1), the reaction mixture is,
Figure QLYQS_11
is the open circuit voltage of the battery>
Figure QLYQS_13
Based on the battery terminal voltage>
Figure QLYQS_16
Ohmic internal resistance of lithium battery, and/or>
Figure QLYQS_18
Is the working current of the battery>
Figure QLYQS_20
And &>
Figure QLYQS_22
Is a polarization resistance, is->
Figure QLYQS_23
And &>
Figure QLYQS_8
Is a polarized capacitor>
Figure QLYQS_10
And &>
Figure QLYQS_12
Is a polarization resistance->
Figure QLYQS_14
And &>
Figure QLYQS_15
The corresponding voltage->
Figure QLYQS_17
And &>
Figure QLYQS_19
Are respectively in>
Figure QLYQS_21
And &>
Figure QLYQS_9
The derivative of (c).
4. The lithium battery SOC estimation method based on weight and volume point self-adaption of claim 1, wherein in the step C, a least square method with forgetting factor recursion is used for conducting equivalent circuit model resistance-capacitance parameters
Figure QLYQS_24
And identifying and obtaining the corresponding resistance-capacitance value.
5. The weight and volume point adaptive-based lithium battery SOC estimation method according to claim 1, wherein in step D, a discrete time state and a measurement state nonlinear equation of a second-order model are established; selecting according to an ampere-hour integral equation (2) and a battery state equation (1)
Figure QLYQS_25
As the state variables, the state equation (3) and the measurement equation (4) of the lithium ion battery can be listed through discretization;
Figure QLYQS_26
(2)/>
Figure QLYQS_27
(3)
Figure QLYQS_28
(4)
in the formulae (2), (3) and (4),
Figure QLYQS_34
and &>
Figure QLYQS_35
SOC values of the battery at the time k and the time k-1, respectively>
Figure QLYQS_37
For the maximum available capacity of the battery, is selected>
Figure QLYQS_40
For coulombic efficiency, is>
Figure QLYQS_41
Is a sampling period, is>
Figure QLYQS_43
Is a time constant->
Figure QLYQS_45
And
Figure QLYQS_29
polarization resistance->
Figure QLYQS_31
The corresponding voltage->
Figure QLYQS_33
And &>
Figure QLYQS_36
Polarization resistance for time k and for time k-1, respectively>
Figure QLYQS_38
The corresponding voltage->
Figure QLYQS_39
Is the open-circuit voltage corresponding to the SOC value of the battery at the moment k>
Figure QLYQS_42
Terminal voltage of the battery at time k->
Figure QLYQS_44
For the operating current of the battery at the moment k>
Figure QLYQS_30
Is process noise->
Figure QLYQS_32
To observe the noise.
6. The weight and volume point adaptive-based lithium battery SOC estimation method according to claim 1, wherein in step E, a weight and volume point adaptive Kalman filter algorithm is used to estimate the SOC of the lithium battery, specifically as follows:
(a) Value of initialized state variable
Figure QLYQS_46
Process noise covariance ≥ v>
Figure QLYQS_47
Measuring noise covariance->
Figure QLYQS_48
And state error covariance
Figure QLYQS_49
；
(b) Calculating a mean of values of state variables
Figure QLYQS_50
And combining the state error covariance>
Figure QLYQS_51
Singular value decomposition is carried out, and cosine similarity is calculated
Figure QLYQS_52
The decomposition and calculation method is as follows:
Figure QLYQS_53
(5)
in the formula
Figure QLYQS_56
Is state error covariance ^ h->
Figure QLYQS_58
Column matrix->
Figure QLYQS_60
Based on the expectation of the value of the state variable>
Figure QLYQS_62
Is the mean value of the state variable>
Figure QLYQS_64
Is state error covariance->
Figure QLYQS_66
Is->
Figure QLYQS_68
Three matrices obtained by singular value decomposition->
Figure QLYQS_54
Is->
Figure QLYQS_57
Is greater than or equal to>
Figure QLYQS_59
Is->
Figure QLYQS_61
Is determined by the feature vector of (a), device for selecting or keeping>
Figure QLYQS_63
Is->
Figure QLYQS_65
Is transferred and is taken out>
Figure QLYQS_67
Is->
Figure QLYQS_69
Transposed matrix of (4), in conjunction with the activation of the key>
Figure QLYQS_55
Is a diagonal matrix;
(c) Determining a height volume criterion by using a high-order radial criterion, and generating a corresponding volume according to the cosine similarity and the height volume criterionAccumulation point
Figure QLYQS_70
And the weight->
Figure QLYQS_71
：
Figure QLYQS_72
(6)
In the formula
Figure QLYQS_74
,/>
Figure QLYQS_76
Is->
Figure QLYQS_78
Is greater than or equal to>
Figure QLYQS_79
A column matrix; />
Figure QLYQS_81
Is the dimension of the state equation, <' > is>
Figure QLYQS_83
Is a variable constant, generally taken to be 1.6, ° v>
Figure QLYQS_85
Is the first->
Figure QLYQS_87
Each volume point is selected and/or judged>
Figure QLYQS_88
Is the first->
Figure QLYQS_90
Each volume point is selected and/or judged>
Figure QLYQS_92
Is the first->
Figure QLYQS_93
Individual volume point, < '> or <' >>
Figure QLYQS_95
Is the first->
Figure QLYQS_96
Each volume point is selected and/or judged>
Figure QLYQS_97
Is the first->
Figure QLYQS_73
The weight of a respective volume point->
Figure QLYQS_75
Is the first->
Figure QLYQS_77
A volume point weight, based on the weight of the volume point>
Figure QLYQS_80
Is the first->
Figure QLYQS_82
Individual volume point weight, <' > based on>
Figure QLYQS_84
Is the first->
Figure QLYQS_86
A volume point weight, based on the weight of the volume point>
Figure QLYQS_89
Is->
Figure QLYQS_91
Is also based on the probability value of>
Figure QLYQS_94
A weight value;
(d) Calculating a state variable predicted value and a state error covariance value;
Figure QLYQS_98
(7)
in the formula
Figure QLYQS_113
Is the state equation, <' >>
Figure QLYQS_115
Is the first->
Figure QLYQS_116
At that moment, is greater or less>
Figure QLYQS_117
Is->
Figure QLYQS_118
A status variable predictor value at the time instant>
Figure QLYQS_119
For an error covariance predictor, <' >>
Figure QLYQS_120
Is->
Figure QLYQS_99
At a moment in time +>
Figure QLYQS_102
Weight of a respective volume point>
Figure QLYQS_104
Is->
Figure QLYQS_106
Is at a moment->
Figure QLYQS_108
Each volume point is selected and/or judged>
Figure QLYQS_110
Is->
Figure QLYQS_112
Transposition of the status variable predictor at a time instant>
Figure QLYQS_114
Is->
Figure QLYQS_100
The time of day process noise covariance matrix, <' >>
Figure QLYQS_101
Is->
Figure QLYQS_103
Is at a moment->
Figure QLYQS_105
Transposition of the state function values of individual volume points,. According to the value of the volume point, the value is greater than or equal to>
Figure QLYQS_107
Is->
Figure QLYQS_109
Is at a moment->
Figure QLYQS_111
The state function values of the individual volume points;
(e) Performing volume point calculation and weight calculation again by using the formula (6) according to the state variable predicted value and the state error covariance predicted value obtained in the step (d);
(f) Updating the measurement predicted value and the measurement autocorrelation and cross-correlation covariance value;
Figure QLYQS_121
(8)
in the formula
Figure QLYQS_138
For the measurement equation, <' >>
Figure QLYQS_139
Is->
Figure QLYQS_140
The measured predicted value at the moment is greater or less than>
Figure QLYQS_141
Is->
Figure QLYQS_142
Is at a moment->
Figure QLYQS_143
The weight of a respective volume point->
Figure QLYQS_144
Is->
Figure QLYQS_122
The moment measured autocorrelation error covariance matrix, <' > is then evaluated>
Figure QLYQS_124
Is->
Figure QLYQS_126
Time measurement cross-correlation error covariance matrix,/>>
Figure QLYQS_127
Is->
Figure QLYQS_129
Is at a moment->
Figure QLYQS_132
Each volume point is selected and/or judged>
Figure QLYQS_134
Is->
Figure QLYQS_137
The transfer of the measured predicted value of the time>
Figure QLYQS_123
Is->
Figure QLYQS_125
A measured noise covariance matrix at a time, based on a time of day>
Figure QLYQS_128
Is->
Figure QLYQS_130
At a moment in time +>
Figure QLYQS_131
The transpose of the measurement function values for each volume point,
Figure QLYQS_133
is->
Figure QLYQS_135
Is at a moment->
Figure QLYQS_136
Measuring function values of the volume points;
(g) Calculating gain
Figure QLYQS_145
Updating the state estimation value and the state covariance estimation value;
Figure QLYQS_146
;
in the formula
Figure QLYQS_148
Is->
Figure QLYQS_149
The voltage data measured by the battery management system is based on the time>
Figure QLYQS_151
Is->
Figure QLYQS_153
Time state error covariance matrix,/>, greater than zero>
Figure QLYQS_156
Is->
Figure QLYQS_158
The moment measured autocorrelation error covariance matrix, <' > is then evaluated>
Figure QLYQS_160
Is the gain matrix at time k +1>
Figure QLYQS_147
Transposing the gain matrix for the moment k +1, superscript @>
Figure QLYQS_150
Stands for transposed, subscript->
Figure QLYQS_152
Represents a fifth->
Figure QLYQS_154
At a moment in time>
Figure QLYQS_155
Is the value of the state variable at the time k +1,
Figure QLYQS_157
is->
Figure QLYQS_159
Moment measurement cross-correlation error covariance matrix, <' >>
Figure QLYQS_161
Is an error covariance predictor;
(h) And (c) repeating the processes from (b) to (g), calculating the state variable and the state covariance at the next moment until the operation is finished, and extracting the result to obtain the SOC estimated value.
CN202310114464.4A 2023-02-15 2023-02-15 Lithium battery SOC estimation method based on weight and volume point self-adaption Active CN115902667B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310114464.4A CN115902667B (en) 2023-02-15 2023-02-15 Lithium battery SOC estimation method based on weight and volume point self-adaption

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310114464.4A CN115902667B (en) 2023-02-15 2023-02-15 Lithium battery SOC estimation method based on weight and volume point self-adaption

Publications (2)

Publication Number Publication Date
CN115902667A true CN115902667A (en) 2023-04-04
CN115902667B CN115902667B (en) 2023-05-23

Family

ID=85735490

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310114464.4A Active CN115902667B (en) 2023-02-15 2023-02-15 Lithium battery SOC estimation method based on weight and volume point self-adaption

Country Status (1)

Country Link
CN (1) CN115902667B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116381513A (en) * 2023-06-07 2023-07-04 广东电网有限责任公司东莞供电局 Lithium ion battery model parameter robust identification method considering measured value abnormality

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1062539A (en) * 1996-08-20 1998-03-06 Japan Marine Sci & Technol Center Generator of sound velocity fluctuation covariance matrix
CN104121907A (en) * 2014-07-30 2014-10-29 杭州电子科技大学 Square root cubature Kalman filter-based aircraft attitude estimation method
CN106352876A (en) * 2016-07-25 2017-01-25 北京航空航天大学 Airborne distributed POS transfer alignment method based on H infinity and CKF hybrid filtering
CN109031276A (en) * 2017-06-09 2018-12-18 浙江工商大学 Adaptive iteration volume kalman filter method in target following with forgetting factor
CN109633479A (en) * 2019-01-11 2019-04-16 武汉理工大学 Lithium battery SOC estimation on line method based on built-in capacitor G-card Kalman Filtering
CN110032812A (en) * 2019-04-18 2019-07-19 河海大学 A kind of dynamic state estimator method based on adaptive volume Kalman filtering
CN110395141A (en) * 2019-06-27 2019-11-01 武汉理工大学 Dynamic lithium battery SOC estimation method based on adaptive Kalman filter method
CN110895146A (en) * 2019-10-19 2020-03-20 山东理工大学 Synchronous positioning and map construction method for mobile robot
CN111537894A (en) * 2020-05-29 2020-08-14 合肥工业大学 Method for estimating SOC and SOP of lithium battery
WO2021201354A1 (en) * 2020-03-30 2021-10-07 중앙대학교 산학협력단 System and method for managing lifespan of electrochemical battery through optimal synthetic surface pressure excitation
CN115170835A (en) * 2022-07-29 2022-10-11 深圳大学 Pedestrian re-identification-based measurement loss frame improvement method and related equipment

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH1062539A (en) * 1996-08-20 1998-03-06 Japan Marine Sci & Technol Center Generator of sound velocity fluctuation covariance matrix
CN104121907A (en) * 2014-07-30 2014-10-29 杭州电子科技大学 Square root cubature Kalman filter-based aircraft attitude estimation method
CN106352876A (en) * 2016-07-25 2017-01-25 北京航空航天大学 Airborne distributed POS transfer alignment method based on H infinity and CKF hybrid filtering
CN109031276A (en) * 2017-06-09 2018-12-18 浙江工商大学 Adaptive iteration volume kalman filter method in target following with forgetting factor
CN109633479A (en) * 2019-01-11 2019-04-16 武汉理工大学 Lithium battery SOC estimation on line method based on built-in capacitor G-card Kalman Filtering
CN110032812A (en) * 2019-04-18 2019-07-19 河海大学 A kind of dynamic state estimator method based on adaptive volume Kalman filtering
CN110395141A (en) * 2019-06-27 2019-11-01 武汉理工大学 Dynamic lithium battery SOC estimation method based on adaptive Kalman filter method
CN110895146A (en) * 2019-10-19 2020-03-20 山东理工大学 Synchronous positioning and map construction method for mobile robot
WO2021201354A1 (en) * 2020-03-30 2021-10-07 중앙대학교 산학협력단 System and method for managing lifespan of electrochemical battery through optimal synthetic surface pressure excitation
CN111537894A (en) * 2020-05-29 2020-08-14 合肥工业大学 Method for estimating SOC and SOP of lithium battery
CN115170835A (en) * 2022-07-29 2022-10-11 深圳大学 Pedestrian re-identification-based measurement loss frame improvement method and related equipment

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
JUNJIE BAI,ET AL.: "ACKF filtering algorithm based on exponential weighting" *
冯亚丽: "非线性卡尔曼滤波算法的改进及精度分析" *
肖俊等: "基于扰动后暂态量的戴维南等值参数辨识方法" *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116381513A (en) * 2023-06-07 2023-07-04 广东电网有限责任公司东莞供电局 Lithium ion battery model parameter robust identification method considering measured value abnormality
CN116381513B (en) * 2023-06-07 2023-07-28 广东电网有限责任公司东莞供电局 Lithium ion battery model parameter robust identification method considering measured value abnormality

Also Published As

Publication number Publication date
CN115902667B (en) 2023-05-23

Similar Documents

Publication Publication Date Title
CN110286332B (en) Electric vehicle power battery SOC estimation method based on multiple innovation theory
CN110688808B (en) Particle swarm and LM optimization hybrid iterative identification method of power battery model
CN111679199B (en) Lithium ion battery SOC estimation method and device
CN111098755B (en) SOC estimation method for power battery of electric vehicle
CN109358293B (en) Lithium ion battery SOC estimation method based on IPF
CN111812515A (en) XGboost model-based lithium ion battery state of charge estimation
CN112083334A (en) Lithium ion battery state of charge estimation method based on data driving
CN112858920B (en) SOC estimation method of all-vanadium redox flow battery fusion model based on adaptive unscented Kalman filtering
CN111426957A (en) SOC estimation optimization method for power battery under simulated vehicle working condition
CN115902667A (en) Lithium battery SOC estimation method based on weight and volume point self-adaption
Takyi-Aninakwa et al. Enhanced multi-state estimation methods for lithium-ion batteries considering temperature uncertainties
CN112528472A (en) Multi-innovation hybrid Kalman filtering and H-infinity filtering algorithm
CN111950205A (en) Lithium battery SOH prediction method based on FWA optimization extreme learning machine
CN114114038A (en) Lithium battery SOC and available capacity joint estimation method under full-life and full-temperature conditions
CN112946481A (en) Based on federation H∞Filtering sliding-mode observer lithium ion battery SOC estimation method and battery management system
CN113128672A (en) Lithium ion battery pack SOH estimation method based on transfer learning algorithm
CN111751750A (en) Multi-stage closed-loop lithium battery SOC estimation method based on fuzzy EKF
CN116754959A (en) SOC estimation method based on improved GWO optimized forgetting factor on-line parameter identification
CN115308608A (en) All-vanadium redox flow battery voltage prediction method, device and medium
Zhang et al. Estimation of Lithium-ion battery state of charge
CN110058162A (en) A kind of parameter identification method based on linearly invariant battery model structure
CN110927585A (en) Lithium battery SOH estimation system and method based on self-circulation correction
CN116718920A (en) Lithium battery SOC estimation method based on RNN (RNN-based optimized extended Kalman filter)
CN115421056A (en) Cross-scale multi-state joint estimation method suitable for super capacitor
CN114089193A (en) Method and device for estimating temperature and negative pole potential of battery on line and computer equipment

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