CN112858918B - Power lithium ion battery health state online estimation method based on optimization multi-factor - Google Patents

Power lithium ion battery health state online estimation method based on optimization multi-factor Download PDF

Info

Publication number
CN112858918B
CN112858918B CN202110059118.1A CN202110059118A CN112858918B CN 112858918 B CN112858918 B CN 112858918B CN 202110059118 A CN202110059118 A CN 202110059118A CN 112858918 B CN112858918 B CN 112858918B
Authority
CN
China
Prior art keywords
battery
charging
health state
factor
health
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
CN202110059118.1A
Other languages
Chinese (zh)
Other versions
CN112858918A (en
Inventor
夏向阳
邓子豪
张嘉诚
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.)
Changsha University of Science and Technology
Original Assignee
Changsha University of Science and Technology
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 Changsha University of Science and Technology filed Critical Changsha University of Science and Technology
Priority to CN202110059118.1A priority Critical patent/CN112858918B/en
Publication of CN112858918A publication Critical patent/CN112858918A/en
Application granted granted Critical
Publication of CN112858918B publication Critical patent/CN112858918B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • 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]
    • G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
    • 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]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health

Abstract

The invention relates toA power lithium ion battery health state online estimation strategy based on multiple factors is optimized, and the problem that the health state online estimation precision of the conventional electric automobile power battery is not high in the charging process is solved. The method mainly comprises the following steps: 1: setting constraint conditions under different factors, dynamically searching and optimizing a charging voltage segment U by using a genetic algorithm with minimum estimation error as an objective function A ’ ‑U B ’ (ii) a 2: in optimizing the charging section U A ’ ‑U B ’ Based on the above, the influence of the three factors of the charging electric quantity, the charging time and the battery internal resistance on the estimation of the battery health state is normalized, and the health state H corresponding to each factor is respectively obtained i1 、H i2 、H i3 .3: optimizing weight coefficient based on least square method and calculating comprehensive health factor H i 。

Description

Power lithium ion battery health state online estimation method based on optimization multi-factor
Technical Field
The invention relates to an optimized multi-factor based online estimation method for the health state of a power lithium ion battery, and belongs to the technical field of electric power.
Background
Accurate assessment of the state of health of a power lithium ion battery is critical to ensuring the safety and reliability of the battery. However, in the actual use of the electric vehicle, the actual working conditions are complex, the discharging process is greatly influenced by user behaviors, the uncertainty is prominent, and the accurate evaluation of the health state of the battery is not facilitated. But the charging phase is controllable and the charging voltage and current data is preserved, so the estimation of the state of health of the battery during the charging process is more reliable.
The existing estimation method is generally divided into a data driving model and an equivalent circuit model, wherein the data driving is based on an empirical model, an Arrhenius capacity aging model and the like, and the estimation precision is not high; the estimation method based on the battery equivalent circuit model mostly uses the ohmic internal resistance as a key technical index for estimating the health state of the battery according to the approximate linear relationship between the ohmic internal resistance and the health state of the battery, but the situations of parameter mismatching and large estimation error easily occur in the long-term health estimation of the power battery. Furthermore, while parameters such as polarization capacitance and polarization impedance may help improve the accuracy of the estimation of battery health, accurate on-board measurement of such parameters requires additional experimentation under specific conditions, which is difficult to achieve in the practical deployment of electric vehicles, and which may cause unnecessary additional damage to the battery.
Disclosure of Invention
In order to improve the estimation precision of the charging process of the power battery of the electric automobile, the invention provides an optimized multi-factor based online estimation method for the health state of the power lithium ion battery.
The method comprises the following basic steps:
step S1: during actual use of the electric vehicle, full charge rarely occurs, considering the user's actual battery charging range, selecting an appropriate voltage segment according to the battery OCV-SOC curve, 85U% N Corresponding SOC of about 30%,95% N The corresponding SOC is about 90%, considering the practicality of the proposed strategy, therefore setting the boundaries of the optimization voltage in advance to 85% U N ～95％U N 。
The estimation accuracy of the internal resistance of the battery under different open-circuit voltage segments is considered to be different, the impedance characteristic change of the battery in the voltage segment corresponding to the SOC of 60% -80% is relatively smooth, the error of the internal resistance estimation is relatively small, the estimation accuracy of the health state of the battery can be improved, the internal chemical reaction of the battery under other voltage segments is more severe, the impedance characteristic change is more unstable, and the estimation of the health state of the battery is also inaccurate.
In addition, if the charging voltage segment is too narrow, the corresponding charged electric quantity is too small, and the robustness of the actual capacity of the battery estimated by the factor is not high; if the charging voltage segment range is too wide, the factor is not well suited for actual vehicle conditions. In summary, the boundary conditions for setting the charging voltage are as follows:
Figure GDA0003848548130000021
after the voltage boundary conditions are determined, the optimal charging segment is found by a genetic algorithm. Firstly, sequencing serial numbers i =1,2,3, n of batteries in a training set, wherein the actual capacity of each battery in the kth charge-discharge cycle is C 1k ,C 2k ,...,C nk The charging voltage segment corresponds to a charging capacity of
Figure GDA0003848548130000022
Because the two have strong linear correlation, the least square method can be used for carrying out linear fitting on the two, and the fitting formula is as follows.
Figure GDA0003848548130000023
In the formula k i ,b i Respectively, the slope and intercept of the linear fit.
Taking the average slope and average intercept of n cells in the training set,
Figure GDA0003848548130000024
calculating an estimated battery capacity of each battery in the training set as
Figure GDA0003848548130000025
Where i =1,2,3.
The root mean square error of the estimated battery capacity of the ith battery in the training set is calculated as
Figure GDA0003848548130000026
The objective function of the genetic algorithm is
Figure GDA0003848548130000027
Figure GDA0003848548130000028
Wherein r is p Denotes the Pearson index, r s The Spearman index is shown and is obtained by the following formula.
Figure GDA0003848548130000031
x represents the input charging time t, y represents the battery state of health H i2 。r p ’,r s ' denotes the correlation index under the optimized voltage segment, and the closer the two indexes are to 1, the stronger the correlation between the two variables is, and the smaller the correlation is vice versa.
Step S2: charging the electric quantity in the optimized voltage segment
Figure GDA0003848548130000032
And battery capacity C i ' has strong homogeneous linear relation, and can further estimate the state of health H of the battery after calculating the battery capacity through the charge capacity i1 。
Figure GDA0003848548130000033
Figure GDA0003848548130000034
Selecting a mean kernel function and a covariance kernel function based on a Gaussian regression (GPR) model trained by a data driving method, and optimizing a charging segment U after initialization setting A ’-U B ' corresponding charging time t i As input, its correlation index r p And r s Closer to 1, the regression value, i.e. the state of health H of the battery, is obtained by regression through the Gaussian process i2 The accuracy of the method is higher, a GPR model is described in detail in many documents at present, a specific regression algorithm is not described in detail, and the optimized battery state of health H is obtained by using the model directly based on the optimization parameters i2 。
Because the battery impedance characteristic change in the optimized voltage segment is relatively smooth and the error of the internal resistance estimation is smaller, the corresponding internal resistance r under the optimized voltage segment can be obtained by reading the data of the vehicle-mounted BMS system 0i Thereby improving the estimation precision of the battery state of healthAnd (4) degree. Calculating the State of health H of the Battery i3 As shown in the following formula.
Figure GDA0003848548130000035
In the formula: r is new The internal resistance of a brand new battery; r is 0i Is the current internal resistance of the battery; r is old The internal resistance of the battery in retirement.
Therefore, the influence of the three factors of the charged electric quantity, the charging time and the internal resistance of the battery under the optimized voltage segment on the estimated battery health state is normalized, and the health state H corresponding to each factor is obtained respectively i1 、H i2 、H i3 。
And step S3: and calculating the comprehensive health state of the battery by optimizing the setting of the weight coefficient based on a least square method. Since the influence of each factor on the state of health of the battery has been normalized in step S2, the integrated state of health H is assumed i Is an independent variable H i1 、H i2 、H i3 Is a linear function of (a).
H i ＝α 1 H i1 +α 2 H i2 +α 3 H i3
Wherein H i1 For optimizing the quantity of electricity C charged in the voltage segment i ' estimated State of health of Battery, H i2 Charging time t required for optimizing voltage segment i Estimated state of health of the battery, H i3 For optimizing the corresponding internal resistance r under the voltage segment 0i An estimated state of health of the battery. Alpha is alpha 1 、α 2 、α 3 Are weight coefficients. The multi-factor model that calculates the i battery health states may be expressed as
Figure GDA0003848548130000041
Is converted into a matrix type,
Figure GDA0003848548130000042
then, the vector is simplified into a vector expression,
H ii ·α＝H
with the minimum estimation error as the target, the objective function is set as follows
Figure GDA0003848548130000043
Based on the principle of least square method, adding Lagrange multiplier constraint condition,
Figure GDA0003848548130000044
wherein, mu j λ is lagrange multiplier, and c is penalty factor. Simplifying to obtain a multiplier iterative formula of
Figure GDA0003848548130000045
Setting the calculation precision, and then the iteration end condition is
Figure GDA0003848548130000046
Wherein epsilon is calculation precision, and when the estimation error reaches minimum, a weight coefficient alpha is obtained 1 、α 2 、α 3 The optimal solution of (1).
Advantageous effects
Compared with the existing method for estimating the health state of the power battery in the charging process, the method takes the minimum estimation error as a target function, sets the multi-factor constraint condition, optimizes the charging voltage segment through a genetic algorithm, improves the influence of three factors, namely the charging electric quantity, the charging time and the internal resistance of the battery, on the accurate estimation of the health state of the battery on the basis, performs normalization processing, and finally comprehensively estimates the health state of the battery through a least square method, so that the overall estimation precision can be improved. And whole estimation process just accomplishes when electric automobile charges, and the practicality is strong, and the easy popularization realizes in each electric automobile charging station.
Drawings
FIG. 1 is a flow chart of policies provided by the present invention
FIG. 2 is a schematic diagram of optimizing voltage slice
Detailed Description
For the convenience of understanding the contents of the embodiments of the present invention, the present invention will be described with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Step S1: setting voltage boundary and different factor constraint conditions, taking minimum estimation error as objective function, and dynamically optimizing charging voltage segment U through genetic algorithm A ’-U B ' the specific optimization procedure is shown in FIG. 2.
Setting boundary conditions
Figure GDA0003848548130000051
Determining an objective function
Figure GDA0003848548130000052
Thereby obtaining an optimized charging voltage segment U A ’-U B ’。
Step S2: in optimizing the charging section U A ’-U B On the basis of the' step, the influence of three factors of the charging electric quantity, the charging time and the internal resistance of the battery on the estimation of the health state of the battery is normalized, and the health state H corresponding to each factor is respectively solved i1 、H i2 、H i3 . The brief procedure is as follows.
Figure GDA0003848548130000053
Gaussian regression (GPR) model trained based on data-driven approach to optimize charging segment U A ’-U B ' corresponding charging time t i For input, the regression value, namely the battery state of health H, is obtained through the regression of a Gaussian process i2 。
Figure GDA0003848548130000061
And step S3: suppose a comprehensive health state H i Is an independent variable H i1 、H i2 、H i3 Is a linear function of (a).
H i ＝α 1 H i1 +α 2 H i2 +α 3 H i3
Conversion to matrix form
H ii ·α＝H
Then based on the least squares method, the objective function is as follows
Figure GDA0003848548130000062
Adding a Lagrange multiplier constraint condition,
Figure GDA0003848548130000063
wherein, mu j λ is lagrange multiplier, and c is penalty factor. Simplifying to obtain a multiplier iterative formula of
Figure GDA0003848548130000064
The iteration end condition is
Figure GDA0003848548130000065
Wherein epsilon is the calculation precision, and the estimation error reaches the maximumHour, obtaining the weight coefficient alpha 1 、α 2 、α 3 The optimal solution of (1).
Finally, the weight coefficient alpha is calculated 1 、α 2 、α 3 Is substituted back to the following formula
H i ＝α 1 H i1 +α 2 H i2 +α 3 H i3
Finally obtaining the comprehensive health state H of the power battery under the optimized charging segment i . The overall flow chart of the optimization-based multi-factor online estimation strategy for battery state of health is shown in fig. 1.
The above description is only an example of the present invention and should not be construed as limiting the present invention, and any modifications, equivalents, improvements and the like made within the spirit of the present invention should be included in the scope of the claims of the present invention.

Claims (1)

1. An optimization multi-factor based online estimation method for the health state of a power lithium ion battery is characterized by comprising the following steps:
step S1: setting voltage boundary and different factor constraint conditions, dynamically searching and optimizing a charging voltage segment U by using a genetic algorithm with minimum estimation error as an objective function A ’-U B ', the boundary conditions for setting the charging voltage are as follows;
Figure FDA0003839507800000011
after the voltage boundary condition is determined, an optimized charging segment is searched through a genetic algorithm, firstly, the serial numbers of the batteries in the training set are ordered, i =1,2,3, n, and the actual capacity of each battery in the kth charging and discharging cycle is C 1k ,C 2k ,...,C nk The charging voltage segment corresponds to a charging capacity of
Figure FDA0003839507800000012
Because of strong linear correlation between the two, the least square method can be used for the twoThe line is fitted linearly, the fitting formula is as follows,
Figure FDA0003839507800000013
in the formula k i ，b i Respectively representing the slope and intercept of the linear fit;
the average slope and average intercept are taken for n cells in the training set, where i =1,2,3, n,
Figure FDA0003839507800000014
calculating the estimated battery capacity of each battery in the training set at the k-th charge-discharge cycle as
Figure FDA0003839507800000015
The root mean square error of the estimated battery capacity of the ith battery in the training set is calculated as
Figure FDA0003839507800000016
The objective function of the genetic algorithm is
Figure FDA0003839507800000017
Figure FDA0003839507800000018
Wherein r is p Denotes the Pearson index, r s Which represents the Spearman index, are respectively obtained by the following formula,
Figure FDA0003839507800000021
x represents the input charging time t, y represents the battery state of health H i2 ，r p ’,r s ' denotes the correlation index under the optimized voltage segment, the closer the two indexes are to 1, the stronger the correlation between the two variables is, and the smaller the correlation is vice versa,
step S2: in optimizing the charging section U A ’-U B On the basis of' the method, the influence of three factors of the charging electric quantity, the charging time and the battery internal resistance on the estimation of the battery health state is normalized, and the health state H corresponding to each factor is respectively solved i1 、H i2 、H i3 ，
And step S3: optimizing weight coefficient based on least square method and calculating comprehensive health factor H i 。
CN202110059118.1A 2021-01-15 2021-01-15 Power lithium ion battery health state online estimation method based on optimization multi-factor Active CN112858918B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110059118.1A CN112858918B (en) 2021-01-15 2021-01-15 Power lithium ion battery health state online estimation method based on optimization multi-factor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110059118.1A CN112858918B (en) 2021-01-15 2021-01-15 Power lithium ion battery health state online estimation method based on optimization multi-factor

Publications (2)

Publication Number Publication Date
CN112858918A CN112858918A (en) 2021-05-28
CN112858918B true CN112858918B (en) 2022-10-28

Family

ID=76005990

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110059118.1A Active CN112858918B (en) 2021-01-15 2021-01-15 Power lithium ion battery health state online estimation method based on optimization multi-factor

Country Status (1)

Country Link
CN (1) CN112858918B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113820615A (en) * 2021-09-30 2021-12-21 国网福建省电力有限公司龙岩供电公司 Battery health degree detection method and device
CN115825743A (en) * 2022-03-07 2023-03-21 宁德时代新能源科技股份有限公司 Battery SOH value calculation method, control device and battery management system

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101459925B1 (en) * 2013-07-05 2014-11-07 현대자동차주식회사 Control method of Low DC/DC Converter for electric vehicle, and Low DC/DC Converter control system using the same
CN107024664B (en) * 2017-04-01 2019-04-12 湖南银杏数据科技有限公司 Lithium battery remaining life quick predict method based on constant-current charge segment
CN107121643B (en) * 2017-07-11 2019-10-11 山东大学 Health state of lithium ion battery combined estimation method
CN107329094A (en) * 2017-08-23 2017-11-07 北京新能源汽车股份有限公司 Electrokinetic cell health status evaluation method and device
CN107831444B (en) * 2017-10-26 2019-12-03 哈尔滨工业大学 A kind of health state of lithium ion battery estimation method
KR20200122628A (en) * 2019-04-18 2020-10-28 현대모비스 주식회사 Method for managing battery for vehicle and apparatus for the same

Also Published As

Publication number Publication date
CN112858918A (en) 2021-05-28

Similar Documents

Publication Publication Date Title
Lai et al. Co-estimation of state of charge and state of power for lithium-ion batteries based on fractional variable-order model
CN110596593B (en) Lithium ion battery SOC estimation method based on intelligent adaptive extended Kalman filtering
CN112858918B (en) Power lithium ion battery health state online estimation method based on optimization multi-factor
CN103616647A (en) Battery remaining capacity estimation method for electric car battery management system
CN107741568B (en) Lithium battery SOC estimation method based on state transition optimization RBF neural network
Huang et al. Soc estimation of li-ion battery based on improved ekf algorithm
CN110824363B (en) Lithium battery SOC and SOE joint estimation method based on improved CKF
CN108872873A (en) A kind of lithium iron phosphate dynamic battery state-of-charge joint estimate method based on GA-AUKF
US20150057957A1 (en) Degradation estimation method for high voltage battery
CN111679199B (en) Lithium ion battery SOC estimation method and device
CN104569835A (en) Method for estimating state of charge of power battery of electric automobile
CN111856178B (en) SOC partition estimation method based on electrochemical characteristics of lithium ion capacitor
CN111308356A (en) SOC estimation method with weighted ampere-hour integration
Li et al. A novel state estimation approach based on adaptive unscented Kalman filter for electric vehicles
CN112327168A (en) XGboost-based electric vehicle battery consumption prediction method
CN113740736A (en) Electric vehicle lithium battery SOH estimation method based on deep network self-adaptation
Jiang et al. A cell-to-pack state estimation extension method based on a multilayer difference model for series-connected battery packs
CN103135066A (en) Measuring method of electric quantity of segmented iron phosphate lithium battery
CN112462282A (en) Method for determining real-time state of charge of battery pack based on mechanism model
Chen et al. A novel battery health indicator and PSO-LSSVR for LiFePO4 battery SOH estimation during constant current charging
CN114861545A (en) Lithium battery SOP online estimation method based on RNN neural network and multi-parameter constraint
CN106707181A (en) Cell parameter and charged state estimation method of lithium ion
Takyi-Aninakwa et al. An ASTSEKF optimizer with nonlinear condition adaptability for accurate SOC estimation of lithium-ion batteries
CN113125967A (en) Lithium battery SOE calculation method based on temperature rise prediction
Wang et al. Research on multiple states joint estimation algorithm for electric vehicles under charge mode

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