CN114839536A - Lithium ion battery health state estimation method based on multiple health factors - Google Patents
Lithium ion battery health state estimation method based on multiple health factors Download PDFInfo
- Publication number
- CN114839536A CN114839536A CN202210352554.2A CN202210352554A CN114839536A CN 114839536 A CN114839536 A CN 114839536A CN 202210352554 A CN202210352554 A CN 202210352554A CN 114839536 A CN114839536 A CN 114839536A
- Authority
- CN
- China
- Prior art keywords
- battery
- charging
- health
- voltage
- model
- 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
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]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/60—Other road transportation technologies with climate change mitigation effect
- Y02T10/70—Energy storage systems for electromobility, e.g. batteries
Abstract
The invention discloses a lithium ion battery health state estimation method based on multiple health factors, belongs to the technical field of battery management, and mainly solves the problem that the estimation accuracy of the battery health state is not high under the condition of quick charging. Based on voltage and current test data in a battery rapid charge-discharge cycle experiment, health factors are extracted from a constant-current charging process to form a characteristic vector, and the characteristic vector comprises charging time, charging energy and information entropy in a local voltage interval in the charging process. And establishing a Gaussian process regression prediction model by taking the feature vector as input and the battery SOH as output, and training the Gaussian process regression model by using experimental data. And in an online state, acquiring an input feature vector, inputting the input feature vector into a trained Gaussian process regression model, and predicting the SOH of the battery. The method does not need to establish a complex battery physical model, can realize the online evaluation of the SOH of the battery by a data driving method, and has very high accuracy and better universality.
Description
Technical Field
The invention relates to a lithium ion battery health state estimation method based on multiple health factors, belongs to the technical field of power battery management, and is used for health state estimation under the condition of rapid charging of a vehicle-mounted power lithium ion battery.
Background
In the field of electric vehicles, lithium ion batteries are the primary energy storage devices. As the battery is used for a longer time, the performance of the battery is degraded, the available capacity and output power of the battery are affected, and safety problems may be caused under certain conditions. The SOH of a battery is represented by the general state of health (SOH) of the battery in the battery industry, and the SOH of a new power battery is regulated to be 1 or 100 percent; and when the performance of the battery can not meet the use requirement, the battery is considered to be invalid, and the service life is over. The SOH values of the battery failures defined in different application occasions are different, and for a pure electric vehicle mainly requiring energy, when the SOH of the power battery is 70%, the battery cannot meet the normal requirement, namely the service life of the battery is finished, and for a hybrid electric vehicle mainly requiring power, the SOH of the power battery is 80% and is often used as the service termination condition of the power battery. Therefore, the lithium ion battery is properly managed and monitored, the health state of the battery is accurately evaluated, the performance of the electric automobile can be guaranteed, abuse of the battery is effectively prevented, and safety accidents are avoided.
The quick charging technology is a more approved entry point for the electric automobile, compared with a common charging mode, the quick charging with large current can save the charging time, and can be compared with the oiling time of the traditional automobile. However, the rapid charging is more likely to cause side reactions and overheating phenomena inside the battery, and has potential damage to the electrode structure of the battery, possibly resulting in rapid change of the state of health of the battery.
Estimation of state of health of a lithium ion battery based on a conventional charging process is generally based on health factors extracted from battery charging experimental data to estimate the state of health of the battery, such as constant current charging time, constant voltage charging time, and the like. The extraction of these factors depends on the conventional constant current constant voltage charging mode. Under the condition of large-current charging, the charging current is usually distributed in a step mode, the maximum current is firstly used, after the limiting voltage is reached, the current is reduced, the charging is continued to the limiting voltage, and the process is repeated. Thus, the charging is segmented and the length of the charging time is closely related to the magnitude of the current used. Under such conditions, the state of health of the battery can no longer be effectively estimated by merely selecting the constant current charging time.
In the practice of estimating the state of health of the battery, due to the inconsistency in the battery, even if the batteries of the same batch size have different performance degradation degrees, even if the battery has a larger difference. The method adopts a plurality of health factors to estimate the health state of the battery, so that the limitation of a single health factor can be effectively solved, and the health state of the battery can be estimated more accurately.
The patent provides an estimation method based on multiple health factors, aiming at the health state estimation of a battery under the current rapid charging condition, an extraction method based on the health factors of partial voltage intervals is designed, and the accurate estimation of the health state of the battery can be realized for the battery which does not use a conventional charging method.
Disclosure of Invention
The invention aims to provide a lithium ion battery health state estimation method based on a quick charging condition, which comprises the following steps: three health factors, namely charging time, charging energy and information entropy in a local voltage interval in the rapid charging process of the battery are extracted, and a Gaussian Process Regression (GPR) model is used for estimating the SOH of the battery. The method can realize the estimation of the battery health state of the battery under the high-current quick charging strategy, and the estimation precision is greatly improved compared with a single health factor.
The specific implementation steps are as follows:
s1, extracting a local voltage interval [ V ] in the large-current charging process a ,V b ]The internal charging time is the first health factor, and can be obtained by using the formula (1)
t＝t b -t a (1)
Wherein, t a For charging voltage up to V a Time corresponding to charging time t b For charging voltage up to V b At the corresponding charging time, t is [ V ] a ,V b ]And charging time corresponding to the voltage interval. The voltage interval can be selected according to the type of the battery, wherein the voltage corresponding to the lithium iron phosphate battery is V a ＝3.15V,V b ＝3.55V。
S2 extracting local voltage interval V in large current charging process a ,V b ]The internal charge energy is a second health factor, and can be obtained by using the formula (2)
Wherein, t a For charging voltage up to V a Time corresponding to charging time t b For charging voltage up to V b At the corresponding charging time, I is the charging current, and v (t) is the time-varying voltage.
S3 extracting partial voltage interval V in large current charging process a ,V b ]The entropy of the internal information is a third health factor, which can be obtained by using the formula (3)
Wherein, the entropy index E k Expressed by a charging voltage distribution within a voltage range defined by a minimum voltage value V a And a maximum voltage value V b And (4) determining. Dividing the voltage range into a fixed number of intervals, i.e. a voltage range [ V ] a ,V b ]Dividing every 0.1V into a small interval, wherein M is the number of the small intervals, and p (i) represents the frequency of the voltage measurement value appearing in each small voltage interval.
S4 using the three health factors extracted above as input and the battery health status as output, establishing training data set and prediction data set
S5 model using Gaussian process regression algorithm, i.e. with HF i And y i Establishing a Gaussian process regression model y as input and output, respectively i ＝f(HF i )+ε i Wherein, in the step (A),
is the health factor, epsilon, extracted in the steps (1), (2) and (3) i Mean 0 and variance σ to obey Gaussian distribution n Expressed as formula (4), y i Is the state of health of the battery at time i. F (HF) i ) Is a function of the health factor, belongs to the Gaussian process and is expressed as
The process is determined by a mean function and a covariance function, expressed as equations (5) and (6), respectively
ε i ～N(0,σ n 2 ) (4)
m HF ＝E(f(HF)) (5)
The mean function and covariance function were chosen to be 0 and Matern5/2, respectively, where Matern5/2 is expressed as equation (7)
Wherein the content of the first and second substances,
σ f and σ l Is a hyper-parameter of the covariance function.
And S6, importing the training data set into a Gaussian process regression model for training, and acquiring and optimizing the hyper-parameters of the model. In the model established in step 5, there is a hyper-parameter Θ ═ σ n ,σ l ,σ f ]And optimizing the hyper-parameters of the model by using the training data to obtain the optimal result. The hyperparameter is optimized using the maximized log marginal likelihood function, as shown in equation (8):
the function comprises three parts, wherein the first part is a data fitting item and represents the fitting degree of the hyper-parameter; the second part is a complexity penalty term which has the function of preventing overfitting; the third part is a constant term. Optimizing the hyper-parameter by adopting a gradient ascending method, and solving a partial derivative of the formula (8) to obtain:
wherein β ═ K HF,HF +σ n 2 I n ) -1 y。
After the optimized hyper-parameters are obtained through the formulas (8) and (9), a new input HF 'is given to the model, and the predicted value y' of the battery state of health is output.
S7, the prediction data set is led into the trained model for verification, and the accuracy of the model is judged according to the root mean square error and the average absolute error.
And S8, under the online condition, the three extracted health factors are used as input vectors of a Gaussian process regression model, and the model outputs the health state of the battery.
Drawings
Fig. 1 shows SOH predicted by single health factor-based GPR model for battery No. 1.
Fig. 2 shows the SOH predicted by the multi-health factor-based GPR model for battery No. 1.
Fig. 3 shows the SOH predicted by the single health factor-based GPR model for battery No. 2.
Fig. 4 shows the SOH predicted by the multi-health factor-based GPR model for battery No. 2.
Fig. 5 shows the predicted SOH of battery No. 3 based on the single health factor GPR model.
Fig. 6 shows the SOH predicted by the multi-health factor-based GPR model for battery No. 3.
Fig. 7 shows the SOH predicted by the single health factor-based GPR model for battery No. 4.
Fig. 8 shows the SOH predicted by multi-health factor-based GPR model for battery No. 4.
Fig. 9 shows the SOH predicted by the single health factor-based GPR model for battery No. 5.
FIG. 10 shows the SOH predicted by the multi-health factor-based GPR model for battery No. 5.
Fig. 11 shows the SOH predicted by the single health factor-based GPR model for battery No. 6.
Fig. 12 shows the SOH predicted by battery No. 6 based on the GPR model for multiple health factors.
FIG. 13 is a schematic flow chart of the method of the present invention.
Detailed Description
The technical scheme of the invention is described in detail in the following with reference to the accompanying drawings and implementation examples.
According to the method, cyclic charge and discharge data of a laboratory lithium iron phosphate battery are used for estimating and calculating SOH of 6 batteries under 3 different charging strategies, and the specifications of the batteries and the charge and discharge working conditions are shown in Table 1.
Table 1 description of charging strategy and discharging test conditions for test cells
The implementation case is as follows:
the batteries of nos. 1 to 6 are respectively subjected to charge-discharge circulation under the working conditions shown in table 1, and the data set is divided into training data and test data based on voltage-current capacity data information obtained by the charge-discharge circulation. Extracting t, E, 1/E from training data k The GPR model is trained with SOH as the output as input to the multivariate GPR model. And in the test data set, the SOH estimation results of the multiple health factors of the 6 batteries are obtained by utilizing the established model. In addition, in order to compare the superiority of the multiple health factor estimation method, the state of health of the battery is estimated by using the charging time t of the battery in the same voltage range as the health factor, and compared with the estimation method using the multiple health factor as an input. Since the Mean Absolute Error (MAE) better reflects the actual condition of the predicted value error, the Root Mean Square Error (RMSE) is very sensitive to the reflection of extra or extra small errors in a group of measurements, and therefore, the accuracy of prediction can be well reflected. The estimation accuracy of both methods is therefore measured in terms of Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), as shown in Table 2.
TABLE 2 SOH estimation error for Single and multiple health factors
From Table 2, the estimation accuracy of the multiple health factors is very high, MAE is in the range of [0.0039,0.0058], and RMSE is in the range of [0.00480.0076 ]. For battery No. 2, the MAE error drops from 0.0193 to 0.0039 and the RMSE drops from 0.0244 to 0.0048. For battery No. 6, the MAE error drops from 0.0149 to 0.0053 and the RMSE drops from 0.0187 to 0.0061. The SOH estimation accuracy of the multiple health factors of the 6 batteries is improved by at least 37% relative to the estimation accuracy of the single health factor, and the estimation accuracy of the rest batteries except the No. 1 and No. 5 batteries is improved by more than 50%, and error analysis shows that the estimation accuracy of the multiple health factors is obviously improved compared with the single health factor.
From the application, the SOH estimation of the battery can be realized by adopting the single-feature estimation method, but the SOH estimation accuracy of the battery is low due to the limitation of the single health feature in reflecting the health state of the battery, and different from the single-feature estimation method, the multi-health-feature estimation method can well make up the limitation defect of the single feature and greatly improve the estimation accuracy. Furthermore, since the health features are extracted based on the partial voltage interval, estimation of the state of health of the battery in the case of incomplete charging can be achieved.
The above examples can effectively demonstrate the superiority of the method of the invention: the method has the advantages that the SOH estimation of the battery under the condition of quick charging is realized by adopting the quick charging experimental data to extract the health factor, the health state of the battery can be accurately estimated under the condition of incomplete charging of the battery, and compared with a single-factor estimation method, the method can make up the defects of a single-feature estimation method and greatly improve the estimation precision. In addition, the method is verified under 6 different batteries and 3 different large-current charging working conditions, and has good universality.
Claims (1)
1. A lithium ion battery health state estimation method based on multiple health factors is characterized in that: the method comprises the steps of charging time t, charging energy E and charging information in a local voltage interval in a rapid charging processEntropy E k Establishing a battery capacity degradation model for the health factor by using a Gaussian process regression algorithm, and finally determining the SOH (state of health) of the battery;
the specific implementation steps are as follows:
step (1): extracting local voltage interval V in large current charging process a ，V b ]The internal charging time is a first health factor and is obtained by using the formula (1):
t＝t b -t a (1)
wherein, t a For charging voltage up to V a Time corresponding to charging time t b For charging voltage up to V b At a charging time t corresponding to [ V ] a ，V b ]Charging time corresponding to the voltage interval;
step (2): extracting local voltage interval V in large current charging process a ，V b ]The internal charging energy is a second health factor, and is obtained by using the formula (2):
wherein, t a For charging voltage up to V a Time corresponding to charging time t b For charging voltage up to V b At the corresponding charging time, I is charging current, and V (t) is time-varying voltage;
and (3): extracting partial voltage interval V in large current charging process a ，V b ]The internal information entropy is a third health factor and is obtained by using a formula (3):
wherein, the entropy index E k Expressed by a charging voltage distribution within a voltage range defined by a minimum voltage value V a And a maximum voltage value V b Determining; dividing the voltage range into a fixed number of intervals, i.e. a voltage range [ V ] a ，V b ]Dividing the voltage into small voltage intervals every 0.1V, wherein M is the number of the small voltage intervals, p (ii) represents the frequency of voltage measurement values appearing in each small voltage interval, and ii is the serial number of the voltage intervals;
and (4): establishing a training data set and a prediction data set by taking the extracted three health factors as input and the health state of the battery as output;
And (5): modeling using a Gaussian process regression algorithm, i.e. with HF i And y i Establishing a Gaussian process regression model y as input and output, respectively i ＝f(HF i )+ε i Wherein, in the step (A),
is the health factor, epsilon, extracted in the steps (1), (2) and (3) i To obey a Gaussian distribution with a mean of 0 and a variance of σ n Expressed as formula (4), y i The state of health of the battery at the moment i; f (HF) i ) Is a function of the health factor, belongs to the Gaussian process and is expressed as
The Gaussian process is determined by a mean function and a covariance function, which are expressed as equations (5) and (6), respectively
ε i ～N(0，σ n 2 ) (4)
m HF ＝E(f(HF)) (5)
The mean function and covariance function were chosen to be 0 and Matem5/2, respectively, where Matern5/2 is expressed as equation (7)
Wherein the content of the first and second substances,
σ f and σ l Is a hyper-parameter of the covariance function;
and (6): importing the training data set into a Gaussian process regression model for training, and acquiring and optimizing hyper-parameters of the model; in the gaussian process regression model established in step (5), the hyperparameter Θ ═ σ n ，σ l ，σ f ]Optimizing the hyper-parameters of the Gaussian process regression model by using the training data to obtain an optimal result; the hyperparameter is optimized using the maximized log marginal likelihood function, as shown in equation (8):
the maximum logarithm marginal likelihood function comprises three parts of contents, wherein the first part is a data fitting item and represents the fitting degree of a hyper-parameter; the second part is a complexity penalty term which has the function of preventing overfitting; the third part is a constant term; optimizing the hyper-parameter by adopting a gradient ascending method, and solving a partial derivative of the formula (8) to obtain:
Wherein β ═ K HF，HF +σ n 2 I n ) -1 y；
Obtaining optimized hyper-parameters through formulas (8) and (9), and then giving a new input HF 'to the model to output a predicted value y' of the battery health state;
and (7): importing the prediction data set into a trained model for verification, and judging the accuracy of the model according to the root mean square error and the average absolute error;
and (8): under the online condition, the three extracted health factors are used as input vectors of the Gaussian process regression model, and the model can output the health state of the battery.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210352554.2A CN114839536A (en) | 2022-04-05 | 2022-04-05 | Lithium ion battery health state estimation method based on multiple health factors |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210352554.2A CN114839536A (en) | 2022-04-05 | 2022-04-05 | Lithium ion battery health state estimation method based on multiple health factors |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114839536A true CN114839536A (en) | 2022-08-02 |
Family
ID=82564193
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210352554.2A Pending CN114839536A (en) | 2022-04-05 | 2022-04-05 | Lithium ion battery health state estimation method based on multiple health factors |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114839536A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116256648A (en) * | 2023-05-16 | 2023-06-13 | 合肥力高动力科技有限公司 | Lithium battery SOH estimation method based on charging data |
| CN116381540A (en) * | 2023-06-05 | 2023-07-04 | 石家庄学院 | Battery health monitoring system under computer running state |
| CN116736141A (en) * | 2023-08-10 | 2023-09-12 | 锦浪科技股份有限公司 | Lithium battery energy storage safety management system and method |
-
2022
- 2022-04-05 CN CN202210352554.2A patent/CN114839536A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116256648A (en) * | 2023-05-16 | 2023-06-13 | 合肥力高动力科技有限公司 | Lithium battery SOH estimation method based on charging data |
| CN116381540A (en) * | 2023-06-05 | 2023-07-04 | 石家庄学院 | Battery health monitoring system under computer running state |
| CN116381540B (en) * | 2023-06-05 | 2023-08-22 | 石家庄学院 | Battery health monitoring system under computer running state |
| CN116736141A (en) * | 2023-08-10 | 2023-09-12 | 锦浪科技股份有限公司 | Lithium battery energy storage safety management system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111707951B (en) | Battery pack consistency evaluation method and system | |
| CN110632528B (en) | Lithium battery SOH estimation method based on internal resistance detection | |
| CN110398697B (en) | Lithium ion health state estimation method based on charging process | |
| CN114839536A (en) | Lithium ion battery health state estimation method based on multiple health factors | |
| CN110146822B (en) | Vehicle power battery capacity online estimation method based on constant-current charging process | |
| CN112034356B (en) | GP-UKF-based online SOH estimation method for electric vehicle power battery | |
| CN107741568B (en) | Lithium battery SOC estimation method based on state transition optimization RBF neural network | |
| CN109633472B (en) | State of charge estimation algorithm of single lithium battery | |
| CN112163372B (en) | SOC estimation method of power battery | |
| CN113109729B (en) | Vehicle power battery SOH evaluation method based on accelerated aging test and real vehicle working condition | |
| Xu et al. | A vehicle-cloud collaborative method for multi-type fault diagnosis of lithium-ion batteries | |
| CN111965560A (en) | Battery health state estimation method oriented to general discharge working condition | |
| CN112904219A (en) | Big data-based power battery health state prediction method | |
| CN112462282B (en) | Method for determining real-time state of charge of battery pack based on mechanism model | |
| CN111983474A (en) | Lithium ion battery life prediction method and system based on capacity decline model | |
| CN110568360A (en) | lithium battery aging diagnosis method based on fuzzy logic algorithm | |
| CN113721159A (en) | Data-driven battery health estimation method oriented to random local charging | |
| CN115754724A (en) | Power battery state of health estimation method suitable for future uncertainty dynamic working condition discharge | |
| CN114545275A (en) | Indirect prediction method for remaining service life of lithium ion battery | |
| CN114646888A (en) | Assessment method and system for capacity attenuation of power battery | |
| CN114462259A (en) | SOC-based battery remaining life prediction method and system, automobile and medium | |
| Wu et al. | State of health estimation for lithium-ion batteries in real-world electric vehicles | |
| CN114184972A (en) | Method and equipment for automatically estimating SOH (state of health) of battery by combining data driving with electrochemical mechanism | |
| CN116466250A (en) | Dynamic working condition model error characteristic-based power battery health state estimation method | |
| CN115097344A (en) | Battery health state terminal cloud collaborative estimation method based on constant voltage charging segments |
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 |