CN115308628A - Battery health state monitoring method - Google Patents
Battery health state monitoring method Download PDFInfo
- Publication number
- CN115308628A CN115308628A CN202210982332.9A CN202210982332A CN115308628A CN 115308628 A CN115308628 A CN 115308628A CN 202210982332 A CN202210982332 A CN 202210982332A CN 115308628 A CN115308628 A CN 115308628A
- Authority
- CN
- China
- Prior art keywords
- lithium ion
- health
- curve
- battery
- ion battery
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000036541 health Effects 0.000 title claims abstract description 90
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012544 monitoring process Methods 0.000 title abstract description 8
- 229910001416 lithium ion Inorganic materials 0.000 claims abstract description 96
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 claims abstract description 88
- 238000002474 experimental method Methods 0.000 claims abstract description 37
- 230000008569 process Effects 0.000 claims abstract description 32
- 230000032683 aging Effects 0.000 claims abstract description 25
- 238000007599 discharging Methods 0.000 claims abstract description 19
- 125000004122 cyclic group Chemical group 0.000 claims abstract description 14
- 238000010219 correlation analysis Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 33
- 238000012360 testing method Methods 0.000 claims description 29
- 230000006870 function Effects 0.000 claims description 22
- 230000015556 catabolic process Effects 0.000 claims description 13
- 238000006731 degradation reaction Methods 0.000 claims description 13
- 239000002245 particle Substances 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000007423 decrease Effects 0.000 claims description 8
- 230000007613 environmental effect Effects 0.000 claims description 7
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 abstract description 3
- WHXSMMKQMYFTQS-UHFFFAOYSA-N Lithium Chemical compound [Li] WHXSMMKQMYFTQS-UHFFFAOYSA-N 0.000 abstract 1
- 229910052744 lithium Inorganic materials 0.000 abstract 1
- 230000000737 periodic effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000004880 explosion Methods 0.000 description 2
- 230000008929 regeneration Effects 0.000 description 2
- 238000011069 regeneration method Methods 0.000 description 2
- 238000007086 side reaction Methods 0.000 description 2
- YZCKVEUIGOORGS-IGMARMGPSA-N Protium Chemical compound [1H] YZCKVEUIGOORGS-IGMARMGPSA-N 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003792 electrolyte Substances 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000047 product Substances 0.000 description 1
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]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
- G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
-
- 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
-
- 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
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention relates to a battery health state monitoring method, which comprises the steps of firstly carrying out cyclic aging experiments on more than 3 lithium electronic batteries respectively to obtain voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process, then extracting health factors reflecting the capacity of each lithium ion battery according to the voltage, current, temperature and capacity data of each lithium ion battery, then selecting 3 health factors with the highest correlation degree from the health factors through grey correlation analysis, finally inputting the 3 health factors into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH estimated value and a confidence interval. The WPSO-GPR model provided by the invention can accurately provide SOH prediction point estimation and 95% probability estimation of the lithium ion battery under variable temperature, and the introduction of an optimization algorithm further improves the prediction performance of the GPR model.
Description
Technical Field
The invention belongs to the technical field of power battery management, and relates to a battery health state monitoring method.
Background
Lithium ion batteries are the core power source for electric vehicles, consumer electronics, and even spacecraft. Therefore, reliability and safety of the lithium ion battery are critical issues in practical applications.
The performance of the battery gradually deteriorates as the service life increases, which may not only affect the proper operation of the electrical equipment, but also have serious consequences. For example, a three-star not 7 battery explosion event, an electric vehicle auto-ignition event, a battery tank explosion event in a partial power plant, etc. that has occurred in recent years. In order to avoid such accidents, SOH (State of health, i.e. percentage of current capacity to factory capacity) of a lithium ion battery has become a hot spot and challenge topic in the Prediction and Health Management (PHM) of electronic products.
Some researchers have replaced the capacity data with indirect features. They can be easily measured in real time and on-line, including current, voltage, temperature, etc. Document 1 (a novel Gaussian process regression model for state-of-health estimation of lithium-ion battery using curve.j. power Sources 2018,384, 387-395) discloses that 4 specific parameters are extracted from the charging curve and used as inputs to the GPR model, instead of the number of cycles. According to the method, only a voltage change curve in the charging process is considered, and the correlation between partial parameters and the capacity is low, so that the prediction accuracy is reduced.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a battery health state monitoring method. The key point of the method is to solve the problems of few samples, low prediction precision, limited applicability and the like in the lithium ion battery health state prediction process, in order to reduce the cost of the monitoring method and improve the prediction precision, the method comprises the steps of measuring lithium ion battery operation data by using a voltage sensor, a temperature sensor, a current sensor and the like, extracting health factors from the collected data, selecting the health factors with high correlation with a capacity degradation curve as high-dimensional input by using a grey correlation analysis method, establishing a GPR (general purpose programming) model for predicting the short-term SOH (state of health) of the lithium ion battery, and providing the monitoring method of the lithium ion battery health state based on different health factors and WPSO-GPR (wavelet-general purpose programming) under different temperature conditions.
In order to achieve the purpose, the invention adopts the following scheme:
a method for predicting the health state of a battery comprises the steps of firstly, carrying out a cyclic aging experiment on M lithium ion batteries respectively to obtain voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process, then extracting health factors reflecting the capacity of each lithium ion battery according to the voltage, current, temperature and capacity data of each lithium ion battery, selecting 3 health factors with the highest correlation degree from the health factors through grey correlation analysis, finally inputting the 3 health factors into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH (state of health) estimation value and a confidence interval;
M≥3。
as a preferred technical scheme:
the method for predicting the state of health of the battery comprises the following specific steps:
(1) Respectively carrying out a cyclic aging experiment on the M lithium ion batteries to obtain voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process; the M lithium ion batteries belong to the same model; controlling the current to be constant in each cyclic aging experiment; the environmental temperature in the cyclic aging experiment of each lithium ion battery is the same, but the environmental temperature in the cyclic aging experiment of different lithium ion batteries is not completely the same;
(2) The following curves for each lithium ion cell were obtained from the voltage, current, temperature and capacity data for each lithium ion cell:
the curve 2 is a voltage curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is voltage;
the curve 3 is a current curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is current;
the curve 4 is a temperature curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is temperature;
(3) Extracting the health factor of each lithium ion battery, and the specific process comprises the following steps: extracting health factors reflecting the capacity of each lithium ion battery from the curves 2, 3 and 4 in the step (2), wherein the health factors are numbered as HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8;
HF1 refers to the time required for the discharge voltage to reach the minimum discharge voltage from the start of discharge per cycle in curve 2;
HF2 refers to the time in curve 4 for the cell temperature to rise from the start of discharge to a maximum value at each cycle;
HF3 refers to the time required for the discharge voltage to decrease from 3.8V to 3.5V per cycle in curve 2;
HF4 refers to the initial maximum slope of the discharge voltage per cycle in curve 2; the calculation method of the initial maximum slope comprises the following steps: dividing a section of curve of which the discharge voltage reaches the lowest discharge voltage from the beginning of discharge in the curve 2 into m sections (the value of m is determined by precision and can be generally 10000), respectively calculating the slope of each section, comparing the slope, and taking the maximum value as the initial maximum slope;
HF5 refers to the maximum slope of the battery temperature at each cycle in curve 4;
HF6 refers to the initial maximum curvature at each cycle in curve 3 (although the control current is constant, during the initial discharge, the current rises from 0 to the controlled current value, "initial" refers to the time calculated for the current to rise from 0 to the controlled current value);
HF7 refers to the discharge time per cycle;
HF8 refers to the area under the current curve at each cycle in curve 3;
(4) Selecting the 3 health factors with the highest correlation from HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8 by grey correlation analysis;
(5) Taking the 3 health factor sequences (HF sequences) with the highest correlation degree obtained in the step (4) and the corresponding capacity sequences as a sample set, wherein the sample set is divided into training samples (x, y) and testing samples (x, y), wherein x is partial data in the health factor sequences, y is partial data in the capacity sequences, x is data in the health factor sequences as predictions, and y is a capacity estimation value as predicted future time;
(6) Optimizing a GPR model by adopting a WPSO algorithm, and then training the GPR model by adopting a training sample to obtain a trained WPSO-GPR model;
(7) And inputting the test sample into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH estimation value and a confidence interval.
In the method for predicting the state of health of the battery, the number of the M lithium ion batteries is 4, the numbers of the M lithium ion batteries are respectively B0005, B0006, B0007 and B0018, and the battery aging test data is acquired from the nasapcae experimental center.
In the above method for predicting the state of health of a battery, the ratio of the training sample to the test sample is 6.
In the method for predicting the state of health of a battery, if an experiment is a single cell experiment, the training sample is defined as a training sample I, and the test sample is defined as a test sample I; performing K-fold cross validation on the total training samples and the total testing samples to screen out training samples I and testing samples I used for predicting the single cell experiment;
if the experiment is a multi-battery experiment, defining the training sample as a training sample II and defining the test sample as a test sample II; in the total training sample and the total test sample, data corresponding to any one of the M lithium ion batteries is used as a test sample II, and data corresponding to other lithium ion batteries is used as a training sample II.
According to the battery state of health prediction method, the SOH estimation error of the trained WPSO-GPR model is within 3%.
According to the battery health state prediction method, the optimization of the GPR model by adopting the WPSO algorithm means that the WPSO algorithm is adopted to automatically search the optimal hyper-parameters in the GPR model;
the formula of the mean function in the GPR model is:
m(x)＝ax+b；
the formula of the covariance function in the GPR model is:
the hyper-parameters are:
wherein a and b are coefficients of a mean function in a GPR model,
is the signal variance,/ 1 ,l 2 Is a variance scale, and p is a periodicity parameter;
in the use process of the lithium ion battery, along with the progress of side reactions in the electrode and the electrolyte, lithium ions are continuously consumed, and the capacity shows a degradation trend. However, the side reaction products may be dissipated during the charging and discharging of the battery, and thus the battery performance may become better and the capacity may be increased in the next cycle period as compared to the previous charging and discharging cycle, which is called capacity regeneration. In order to achieve a more accurate life prediction analysis, two different capacity variation behaviors of the lithium ion battery need to be taken into account. Therefore, for the gaussian process regression model, only a single form of covariance function cannot meet the prediction requirement, and a plurality of covariance functions need to be considered for combined use. The covariance functions can be added or multiplied to construct a combined covariance function that describes the complex problem, subject to non-negativity requirements. Approximately regarding the capacity regeneration phenomenon as periodic variation on the normal capacity degradation trend, so that a periodic covariance kernel function can be selected to describe local variation on the basis of the periodic covariance kernel function, and the periodic covariance function and the squared exponential covariance function are added to be used as a combined covariance function so as to have the generalization capability and the local learning capability at the same time;
the formula in the WPSO algorithm is as follows:
wherein w is the inertial weight; m is the current iteration number; c. C 1 And c 2 Is a non-negative constant as a learning factor, wherein c 1 Is an individual learning factor, c 2 Is a population learning factor; the general values are: c. C 1 ＝2，c 2 ＝2；z i And v i Respectively representing the current position and the speed of the ith particle; r is 1 And r 2 Is distributed in [0,1]]A random number in between; p is i Is the optimal solution for the particle itself, and G is the optimal solution found in the entire population.
In the method for predicting the state of health of the battery, after the WPSO algorithm initializes the particle swarm, the velocity and the position of the particles are updated by the formula in the WPSO algorithm:
in the above method for predicting the state of health of the battery, the value range [ w ] of the inertial weight w min ,w max ]＝[0.4,0.9]。
The battery health state prediction method comprises the following specific steps of (4):
(4.1) one lithium ion battery in the step (1) is selected optionally, 8 health factors extracted from the lithium ion battery in the step (3) are used as 8 groups of comparison sequences, and a reference sequence is determined according to the data in the curve 1 in the step (2);
the expression for noting the jth health factor as the jth comparison sequence is:
HF j ＝[hf j (1),hf j (2),…,hf j (k),…,hf j (n)]；
let the expression of the reference sequence be:
HF 0 ＝[hf 0 (1),hf 0 (2),…,hf 0 (k),…,hf 0 (n)]；
wherein HF is j J is the serial number of the health factor and j =1,2, \8230, 8, k is the cycle number and k =1,2, \8230, n, n is the maximum cycle number of the selected lithium ion battery in the cycle aging experiment; hf (h) f j (k) The data corresponding to the k cycle in the jth health factor is obtained; HF (high frequency) 0 Reference sequence, hf, for the cell capacity of the lithium-ion cell in curve 1 0 (k) The battery capacity corresponding to the kth cycle in curve 1;
the maximum cycle number refers to the cycle number corresponding to the discharge voltage of the lithium ion battery when the discharge voltage reaches a cut-off voltage in a cycle aging experiment;
the cut-off voltage, also called the end voltage, is the lowest working voltage value that the battery is not suitable for discharging again when the battery discharges, and the data is provided by the manufacturer;
(4.2) calculating the difference value of the position elements corresponding to each comparison sequence and the reference sequence one by one, and determining the minimum difference value and the maximum difference value of the position elements corresponding to the comparison sequence and the reference sequence;
the minimum difference is calculated as:
the maximum difference is calculated as:
(4.3) calculating the association coefficient of each comparison sequence and the reference sequence in each cycle, wherein the association coefficient xi of the jth comparison sequence j (k) The calculation formula of (a) is as follows:
in the formula, rho is a resolution coefficient, rho belongs to [0,1], and generally takes a value of 0.5;
(4.4) determining a gray correlation γ of each health factor with the capacity degradation curve j Grey correlation analysis is used to analyze the similarity of the trends in the comparison and reference sequences. Namely, the more similar the geometrical shapes of the curves are, the greater the correlation degree among the factors is, and the closer the gray correlation degree is to 1; conversely, the smaller the correlation between the factors, the gray correlation γ j The calculation formula of (c) is as follows:
(4.5) correlation to Gray j Is ordered by the gray degree of association gamma j And taking the health factor corresponding to the maximum 3 grey correlation degrees as the 3 health factors with the highest correlation degree.
The principle of the invention is as follows:
the discharge process of the lithium ion battery plays an important role in predicting the aging life of the battery. In order to overcome the problem of non-measurable capacity, the invention extracts a measurable degradation index which is highly related to capacity degradation from the discharge process of the lithium ion battery. The main contribution of the method is that the method can predict the SOH of the lithium ion battery by using the indirect health index and the GPR model. Firstly, in order to reduce the cost of a prediction method and improve the prediction precision, the invention provides that indirect health indexes such as voltage, current and temperature curves are extracted from data acquired by a voltage sensor, a temperature sensor, a current sensor and the like; and then selecting a health factor with high correlation with a capacity degradation curve as high-dimensional input through grey correlation analysis, and establishing a WPSO-GPR model to predict the short-term SOH of the lithium ion battery.
Advantageous effects
The WPSO-GPR model provided by the invention can accurately provide SOH prediction point estimation and 95% probability estimation of the lithium ion battery under variable temperature, and the introduction of an optimization algorithm further improves the prediction performance of the GPR model. It is worth mentioning that single cell and multi-cell experiments verify the practicability of the method under different working conditions, and the WPSO-GPR can be suitable for regression modeling of a small sample training set, so that the method provided by the invention has strong robustness and generalization capability.
Drawings
Fig. 1 is a graph showing degradation curves of battery capacities of B0005, B0006, B0007, and B0018;
fig. 2 is a schematic diagram of a discharge process voltage curve of the lithium ion battery with the number of B0005 under different cycle times; wherein Cycle150 means the number of cycles 150;
fig. 3 is a schematic diagram of a discharge process temperature curve of the lithium ion battery with the number of B0005 under different cycle times; wherein Cycle150 refers to the number of cycles 150;
fig. 4 is a schematic diagram of a current curve of the lithium ion battery numbered B0005 in the discharging process of the lithium ion battery at different cycle times; wherein Cycle150 refers to the number of cycles 150;
FIG. 5 is a schematic flow chart of the improved WPSO-GPR model for predicting SOH of the battery;
FIG. 6 is a graph of the prediction result of the algorithm and an error graph.
Detailed Description
The present invention will be further described with reference to the following embodiments. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention can be made by those skilled in the art after reading the teaching of the present invention, and these equivalents also fall within the scope of the claims appended to the present application.
The battery aging test data of the invention is collected from NASA Pco E experiment center.
A battery state of health monitoring method, its step is as follows:
(1) Selecting 4 lithium ion batteries (numbered as B0005, B0006, B0007 and B0018) to respectively perform a cyclic aging experiment, and acquiring voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process;
wherein, 4 lithium ion batteries belong to the same model; controlling the current to be constant in each cyclic aging experiment; the environmental temperature in the cycle aging experiment of each lithium ion battery is the same, but the environmental temperatures in the cycle aging experiments of different lithium ion batteries are not completely the same (generally, 3 environmental temperatures are suggested) to obtain sample data at different temperatures;
(2) The following curves for each lithium ion cell were obtained from the voltage, current, temperature and capacity data for each lithium ion cell:
as shown in fig. 1, curve 1 is a degradation curve of the battery capacity of each lithium ion battery, the abscissa is the number of cycles, and the ordinate is the battery capacity;
curve 4 is a temperature curve of the discharging process of the lithium ion battery under different cycle times, the abscissa is time, the ordinate is temperature, taking B0005 as an example, the temperature curve of the discharging process of the lithium ion battery under different cycle times of B0005 is shown in fig. 3;
(3) Extracting the health factor of each lithium ion battery, and the specific process comprises the following steps: extracting health factors reflecting the capacity of each lithium ion battery from the curve 2, the curve 3 and the curve 4 in the step (2), wherein the health factors are numbered as HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8; wherein,
HF1 refers to the time required for the discharge voltage to reach the minimum discharge voltage from the start of discharge per cycle in curve 2; as can be seen from fig. 2, this time decreases as the number of cycles increases;
HF2 refers to the time in curve 4 for the cell temperature to rise from the time discharge begins to a maximum value at each cycle; as can be seen from fig. 3, this time decreases as the number of cycles increases;
HF3 refers to the time required for the discharge voltage to decrease from 3.8V to 3.5V (the time decrease is relatively gradual, not a sharp decrease) per cycle in curve 2; as can be seen from fig. 2, this time decreases as the number of cycles increases;
HF4 refers to the initial maximum slope of the discharge voltage per cycle in curve 2; the calculation method of the initial maximum slope comprises the following steps: dividing a section of curve of the discharge voltage in each cycle in the curve 2 from the beginning of discharge to the lowest discharge voltage into m sections (the value of m is determined by precision and can be generally 10000), respectively calculating the slope of each section, and comparing, wherein the maximum value is taken as the initial maximum slope.
HF5 refers to the maximum slope of the battery temperature at each cycle in curve 4 (i.e., the rate at which the battery temperature rises fastest);
HF6 refers to the initial maximum curvature at each cycle in curve 3 (although the control current is constant, during the initial discharge, the current rises from 0 to the controlled current value, "initial" refers to the time calculated for the current to rise from 0 to the controlled current value);
HF7 refers to the discharge time per cycle (directly recorded by the sensor employed in the cyclic aging experiment);
HF8 refers to the area under the current curve at each cycle in curve 3;
(4) The 3 health factors with the highest correlation degree are selected from HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8 by grey correlation degree analysis, and the specific steps are as follows:
(4.1) one lithium ion battery in the step (1) is selected optionally, 8 health factors extracted from the lithium ion battery in the step (3) are used as 8 groups of comparison sequences, and a reference sequence is determined according to the data in the curve 1 in the step (2);
the 8 sets of comparison sequences are organized as the following matrix:
the expression for marking the jth health factor as the jth comparison sequence is as follows:
HF j ＝[hf j (1),hf j (2),…,hf j (k),…,hf j (n)]；
wherein HF is j J is the serial number of the health factor and j =1,2, \8230, 8, k is the cycle number and k =1,2, \8230, n, n is the maximum cycle number of the selected lithium ion battery in the cycle aging experiment; hf j (k) The data corresponding to the k cycle in the jth health factor is obtained;
the maximum cycle number is the cycle number corresponding to the discharge voltage of the lithium ion battery when the discharge voltage reaches the cut-off voltage in the cycle aging experiment;
the cut-off voltage, also called the end voltage, is the lowest working voltage value that the battery is not suitable for discharging again when the battery discharges, and the data is provided by the manufacturer;
let the expression of the reference sequence be:
HF 0 ＝[hf 0 (1),hf 0 (2),…,hf 0 (k),…,hf 0 (n)]；
wherein, HF 0 Reference sequence, hf, formed for the battery capacity of the lithium-ion battery in curve 1 0 (k) The battery capacity corresponding to the kth cycle in curve 1;
(4.2) calculating the difference value of the position elements corresponding to each comparison sequence and the reference sequence one by one, and determining the minimum difference value and the maximum difference value of the position elements corresponding to the comparison sequence and the reference sequence;
the minimum difference is calculated as:
the maximum difference is calculated as:
(4.3) calculating the association coefficient of each comparison sequence and the reference sequence in each cycle, wherein the association coefficient xi of the jth comparison sequence j (k) The calculation formula of (a) is as follows:
in the formula, rho is a resolution coefficient, rho belongs to [0,1], and generally takes a value of 0.5;
(4.4) determining a gray correlation γ of each health factor with the capacity degradation curve j The calculation formula is as follows:
(4.5) correlation degree γ to Gray j Is ordered by the gray degree of association gamma j The health factor corresponding to the maximum 3 gray correlation degrees is taken as the 3 health factors with the highest correlation degree;
(5) Taking the 3 health factor sequences (HF sequences) with the highest correlation degree obtained in the step (4) and the corresponding capacity sequences as a sample set, wherein the sample set is divided into training samples (x, y) and testing samples (x, y), wherein x is partial data in the health factor sequences, y is partial data in the capacity sequences, x is data in the health factor sequences as predictions, and y is a capacity estimation value as predicted future time;
the ratio of the training sample to the test sample is 6;
(6) Optimizing a GPR model by adopting a WPSO algorithm, and then training the GPR model by adopting a training sample to obtain a trained WPSO-GPR model;
the method for optimizing the GPR model by adopting the WPSO algorithm is characterized in that the WPSO algorithm is adopted to automatically search the optimal hyper-parameters in the GPR model;
the formula of the mean function in the GPR model is:
m(x)＝ax+b；
the formula of the covariance function in the GPR model is:
the hyper-parameters are as follows:
wherein a and b are coefficients of a mean function in a GPR model,
is the signal variance, l 1 ,l 2 Is a variance scale, and p is a periodicity parameter;
the formula in the WPSO algorithm is as follows:
wherein w is the inertial weight; m is the current iteration number; c. C 1 And c 2 Is a non-negative constant for the learning factor, wherein c 1 Is an individual learning factor, c 2 Is a population learning factor; the general values are: c. C 1 ＝2，c 2 ＝2；z i And v i Respectively representing the current position and the speed of the ith particle; r is a radical of hydrogen 1 And r 2 Is distributed in [0,1]]A random number in between; p is i The optimal solution of the particles, G is the optimal solution found in the whole population;
after initializing the particle swarm by the WPSO algorithm, updating the particle speed and position by using a formula in the WPSO algorithm:
G m ＝argminf(G m ),f(P i m+1 )；
the invention takes the particle swarm size as 24, the iteration step number as 100, and the value range [ w ] of the inertia weight w min ,w max ]＝[0.4,0.9]；
(7) Inputting the test sample into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH estimation value and a confidence interval;
as shown in fig. 5, after inputting the training samples and the test samples, the operating process of the GPR model is:
(a) Selecting a covariance function;
(b) Selecting a hyper-parameter according to the covariance function, and setting a hyper-parameter initial value;
(c) Utilizing the optimized hyper-parameter value of the WPSO algorithm;
(d) Training by using a WPSO-GPR model;
(e) Utilizing the prediction of the WPSO-GPR model to output an SOH estimated value and a confidence interval;
(f) Carrying out error analysis on the SOH estimated value;
(g) And (6) ending.
A single cell experiment was performed using the cycle charge/discharge capacity and IHF data of one cell (B0005). 60% of the training set was used in the experiment; the WPSO-GPR model selects a covariance function as a rational secondary covariance, parameters of the covariance function are optimized by using a WPSO algorithm, and the circulating charge-discharge capacity and the IHF data of one battery at different temperatures are adopted to perform a single cell experiment.
The experimental results and relative errors of the 4-fold cross validation experiment at room temperature in the 60% training set-40% testing set are shown in fig. 6, and the SOH estimation errors are 3% except for individual points and mostly 1.5%, reflecting the effectiveness of the constructed HF and the good regression performance of the WPSO-GPR model in lithium ion battery SOH prediction.
Claims (10)
1. A method for predicting the state of health of a battery is characterized in that: firstly, respectively carrying out cyclic aging experiments on M lithium ion batteries to obtain voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process, then extracting health factors reflecting the capacity of each lithium ion battery according to the voltage, current, temperature and capacity data of each lithium ion battery, selecting 3 health factors with the highest correlation degree from the health factors through grey correlation analysis, finally inputting the 3 health factors into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH estimated value and a confidence interval;
M≥3。
2. the battery state of health prediction method of claim 1, characterized in that the specific steps are as follows:
(1) Respectively carrying out a cyclic aging experiment on the M lithium ion batteries to obtain voltage, current, temperature and capacity data of each lithium ion battery in the charging and discharging process; the M lithium ion batteries belong to the same model; controlling the current to be constant in each cyclic aging experiment; the environmental temperature in the cyclic aging experiment of each lithium ion battery is the same, but the environmental temperature in the cyclic aging experiment of different lithium ion batteries is not completely the same;
(2) The following curves for each lithium ion cell were obtained from the voltage, current, temperature and capacity data for each lithium ion cell:
curve 1 is a degradation curve of the battery capacity of each lithium ion battery, the abscissa is the cycle number, and the ordinate is the battery capacity;
the curve 2 is a voltage curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is voltage;
the curve 3 is a current curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is current;
the curve 4 is a temperature curve of the lithium ion battery in the discharging process under different cycle times, the abscissa is time, and the ordinate is temperature;
(3) Extracting the health factor of each lithium ion battery, wherein the specific process comprises the following steps: extracting health factors reflecting the capacity of each lithium ion battery from the curves 2, 3 and 4 in the step (2), wherein the health factors are numbered as HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8;
HF1 refers to the time required for the discharge voltage to reach the minimum discharge voltage from the start of discharge per cycle in curve 2;
HF2 refers to the time in curve 4 for the cell temperature to rise from the start of discharge to a maximum value at each cycle;
HF3 refers to the time required for the discharge voltage to decrease from 3.8V to 3.5V per cycle in curve 2;
HF4 refers to the initial maximum slope of the discharge voltage per cycle in curve 2;
HF5 refers to the maximum slope of the battery temperature at each cycle in curve 4;
HF6 refers to the initial maximum curvature at each cycle in curve 3;
HF7 refers to the discharge time per cycle;
HF8 refers to the area under the current curve at each cycle in curve 3;
(4) Selecting the 3 health factors with the highest correlation from HF1, HF2, HF3, HF4, HF5, HF6, HF7 and HF8 by grey correlation analysis;
(5) Taking the 3 health factor sequences with the highest correlation degree obtained in the step (4) and the corresponding capacity sequences as a sample set, wherein the sample set is divided into training samples (x, y) and testing samples (x, y), x is partial data in the health factor sequences, y is partial data in the capacity sequences, x is data in the health factor sequences as predictions, and y is a capacity estimation value at the predicted future time;
(6) Optimizing a GPR model by adopting a WPSO algorithm, and then training the GPR model by adopting a training sample to obtain a trained WPSO-GPR model;
(7) And inputting the test sample into a trained WPSO-GPR model for prediction, and outputting to obtain an SOH estimated value and a confidence interval.
3. The battery state of health prediction method of claim 2, wherein M is 4.
4. The method of claim 2, wherein the ratio of the training samples to the test samples is 6.
5. The method according to claim 2, wherein if the experiment is a single cell experiment, the training sample is defined as a training sample I, and the test sample is defined as a test sample I; performing K-fold cross validation on the total training sample and the total testing sample to screen out a training sample I and a testing sample I for predicting a single cell experiment;
if the experiment is a multi-battery experiment, defining the training sample as a training sample II and defining the test sample as a test sample II; in the total training sample and the total test sample, data corresponding to any one of the M lithium ion batteries is used as a test sample II, and data corresponding to other lithium ion batteries is used as a training sample II.
6. The battery state of health prediction method of claim 2, wherein the trained WPSO-GPR model has an SOH estimation error within 3%.
7. The battery state of health prediction method of claim 2, wherein optimizing the GPR model using the WPSO algorithm means automatically searching for optimal hyper-parameters in the GPR model using the WPSO algorithm;
the formula of the mean function in the GPR model is:
m(x)＝ax+b；
the formula of the covariance function in the GPR model is:
the hyper-parameters are:
wherein a and b are coefficients of a mean function in a GPR model,
is the signal variance,/ 1 ,l 2 Is a variance scale, and p is a periodicity parameter;
the formula in the WPSO algorithm is as follows:
wherein w is the inertial weight; m is the current iteration number; c. C 1 And c 2 Is a learning factor, wherein, c 1 Is an individual learning factor, c 2 Is a population learning factor; z is a radical of i And v i Respectively representing the current position and the speed of the ith particle; r is 1 And r 2 Are distributed in [0,1]]A random number in between; p i Is the optimal solution for the particle itself, and G is the optimal solution found in the entire population.
8. The battery state of health prediction method of claim 7, wherein after the WPSO algorithm initializes the population of particles, the velocity and position of the particles are updated using the formula in the WPSO algorithm:
G m ＝argminf(G m ),f(P i m+1 )。
9. the battery state of health prediction method of claim 7Characterized by a value range [ w ] of the inertial weight w min ,w max ]＝[0.4,0.9]。
10. The method for predicting the state of health of a battery according to claim 2, wherein the step (4) comprises the following steps:
(4.1) one lithium ion battery in the step (1) is selected optionally, 8 health factors extracted from the lithium ion battery in the step (3) are used as 8 groups of comparison sequences, and a reference sequence is determined according to the data in the curve 1 in the step (2);
the expression for noting the jth health factor as the jth comparison sequence is:
HF j ＝[hf j (1),hf j (2),…,hf j (j),…,hf j (n)]；
let the expression of the reference sequence be:
HF 0 ＝[hf 0 (1),hf 0 (2),…,hf 0 (k),…,hf 0 (n)]；
wherein, HF j J is the serial number of the health factor and j =1,2, \8230, 8, k is the cycle number and k =1,2, \8230, n, n is the maximum cycle number of the selected lithium ion battery in the cycle aging experiment; hf (h) f j (k) The data corresponding to the kth circulation in the jth health factor; HF (high frequency) 0 Reference sequence, hf, for the cell capacity of the lithium-ion cell in curve 1 0 (k) Is the corresponding battery capacity at the kth cycle in curve 1;
the maximum cycle number refers to the cycle number corresponding to the discharge voltage of the lithium ion battery when the discharge voltage reaches a cut-off voltage in a cycle aging experiment;
(4.2) calculating the difference value of the position elements corresponding to each comparison sequence and the reference sequence one by one, and determining the minimum difference value and the maximum difference value of the position elements corresponding to the comparison sequence and the reference sequence;
the minimum difference is calculated as:
the maximum difference is calculated as:
(4.3) calculating the association coefficient of each comparison sequence and the reference sequence in each cycle, wherein the association coefficient xi of the jth comparison sequence j (k) The calculation formula of (c) is as follows:
in the formula, rho is a resolution coefficient and belongs to [0,1];
(4.4) determining a Grey correlation degree gamma of each health factor with the capacity degradation curve j The calculation formula is as follows:
(4.5) correlation to Gray j Is ordered by the gray degree of association gamma j And taking the health factor corresponding to the maximum 3 gray correlation degrees as the 3 health factors with the highest correlation degrees.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210982332.9A CN115308628A (en) | 2022-08-16 | 2022-08-16 | Battery health state monitoring method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210982332.9A CN115308628A (en) | 2022-08-16 | 2022-08-16 | Battery health state monitoring method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115308628A true CN115308628A (en) | 2022-11-08 |
Family
ID=83862055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210982332.9A Pending CN115308628A (en) | 2022-08-16 | 2022-08-16 | Battery health state monitoring method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115308628A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115542186A (en) * | 2022-11-30 | 2022-12-30 | 中国电力科学研究院有限公司 | Method, device, equipment and medium for evaluating state and consistency of energy storage battery |
| CN116381536A (en) * | 2023-03-07 | 2023-07-04 | 华中科技大学 | Regression element learning-based lithium battery health state prediction method and system |
| CN116774075A (en) * | 2023-08-28 | 2023-09-19 | 清华四川能源互联网研究院 | Lithium ion battery health state evaluation method and system |
| CN117607723A (en) * | 2023-11-24 | 2024-02-27 | 广东电网有限责任公司 | Battery health state prediction method, device, equipment and medium |
| CN118504426A (en) * | 2024-07-17 | 2024-08-16 | 南通理工学院 | Electricity consumption management device of energy storage equipment |
-
2022
- 2022-08-16 CN CN202210982332.9A patent/CN115308628A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115542186A (en) * | 2022-11-30 | 2022-12-30 | 中国电力科学研究院有限公司 | Method, device, equipment and medium for evaluating state and consistency of energy storage battery |
| CN115542186B (en) * | 2022-11-30 | 2023-03-14 | 中国电力科学研究院有限公司 | Method, device, equipment and medium for evaluating state and consistency of energy storage battery |
| CN116381536A (en) * | 2023-03-07 | 2023-07-04 | 华中科技大学 | Regression element learning-based lithium battery health state prediction method and system |
| CN116381536B (en) * | 2023-03-07 | 2024-03-19 | 华中科技大学 | Regression element learning-based lithium battery health state prediction method and system |
| CN116774075A (en) * | 2023-08-28 | 2023-09-19 | 清华四川能源互联网研究院 | Lithium ion battery health state evaluation method and system |
| CN117607723A (en) * | 2023-11-24 | 2024-02-27 | 广东电网有限责任公司 | Battery health state prediction method, device, equipment and medium |
| CN118504426A (en) * | 2024-07-17 | 2024-08-16 | 南通理工学院 | Electricity consumption management device of energy storage equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115308628A (en) | Battery health state monitoring method | |
| CN108896914B (en) | Gradient lifting tree modeling and prediction method for health condition of lithium battery | |
| Oji et al. | Data-driven methods for battery soh estimation: Survey and a critical analysis | |
| CN106446940B (en) | A kind of prediction technique of the supercapacitor capacitance degradation trend based on support vector machines | |
| CN113740736B (en) | Electric vehicle lithium battery SOH estimation method based on deep network self-adaption | |
| CN111443293A (en) | Lithium battery state of health (SOH) estimation method based on data driving | |
| CN114325450A (en) | Lithium ion battery health state prediction method based on CNN-BilSTM-AT hybrid model | |
| CN103954913A (en) | Predication method of electric vehicle power battery service life | |
| CN113917334B (en) | Battery health state estimation method based on evolution LSTM self-encoder | |
| Che et al. | Battery health prognostic with sensor-free differential temperature voltammetry reconstruction and capacity estimation based on multi-domain adaptation | |
| Liu et al. | An improved method of state of health prediction for lithium batteries considering different temperature | |
| CN111983474A (en) | Lithium ion battery life prediction method and system based on capacity decline model | |
| CN114966436A (en) | Lithium battery state of charge prediction method, device, equipment and readable storage medium | |
| CN114839538A (en) | Method for extracting degradation characteristics of lithium ion battery for estimating residual life | |
| CN115994441A (en) | Big data cloud platform online battery life prediction method based on mechanism information | |
| CN117825965A (en) | State prediction method and system for lithium ion battery | |
| CN116125306A (en) | Power battery thermal runaway diagnosis method and system based on hybrid prediction framework | |
| Eleftheriadis et al. | Comparative study of machine learning techniques for the state of health estimation of li-ion batteries | |
| Zhang et al. | Online state-of-health estimation for the lithium-ion battery based on an LSTM neural network with attention mechanism | |
| CN118050653A (en) | Method and device for predicting the battery of a diagnostic device using a multi-element converter model | |
| CN117630673A (en) | Method and device for initially providing an aging state model for an energy store | |
| CN116653608A (en) | Electric automobile charging protection and control method, device and storage medium | |
| CN115993537A (en) | Lithium battery capacity prediction method based on correlation analysis and WOA-LSTM | |
| CN115684972A (en) | Lithium ion battery SOH estimation method based on SSA-SVR model | |
| Crocioni et al. | Li-Ion Batteries Releasable Capacity Estimation with Neural Networks on Intelligent IoT Microcontrollers |
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 |