CN112100902B - Lithium ion battery life prediction method based on flow data - Google Patents
Lithium ion battery life prediction method based on flow data Download PDFInfo
- Publication number
- CN112100902B CN112100902B CN202010797577.5A CN202010797577A CN112100902B CN 112100902 B CN112100902 B CN 112100902B CN 202010797577 A CN202010797577 A CN 202010797577A CN 112100902 B CN112100902 B CN 112100902B
- Authority
- CN
- China
- Prior art keywords
- battery
- random forest
- data
- life prediction
- forest 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 26
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 title claims abstract description 23
- 229910001416 lithium ion Inorganic materials 0.000 title claims abstract description 23
- 238000007637 random forest analysis Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000012935 Averaging Methods 0.000 claims abstract description 3
- 238000003066 decision tree Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 description 8
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 6
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 6
- 239000002245 particle Substances 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000005653 Brownian motion process Effects 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005537 brownian motion Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013501 data transformation Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Secondary Cells (AREA)
- Tests Of Electric Status Of Batteries (AREA)
Abstract
The invention discloses a lithium ion battery life prediction method based on stream data, which comprises the following steps: acquiring stream data of battery characteristics, simultaneously constructing a plurality of random forest models, and training the random forest models by utilizing the stream data of the battery characteristics; obtaining a plurality of time series data of battery characteristics to be predicted, wherein one time series data corresponds to one random forest model, then inputting each time series data into the corresponding random forest model for life prediction, averaging life prediction results output by each random forest model, and finally taking the averaged results as final battery life prediction results.
Description
Technical Field
The invention relates to a lithium ion battery life prediction method, in particular to a lithium ion battery life prediction method based on flow data.
Background
The Remaining Useful Life (RUL) of a battery is always a problem throughout from the production to the use of a lithium ion battery. After the lithium ion battery is produced, unqualified batteries with too low life expectancy need to be screened out, and batteries meeting the standards can be classified according to the life expectancy. Therefore, early classification of lithium ion battery life predicts a major focus of RUL prediction.
Currently, there have been many studies on life prediction of lithium ion batteries. Some research has focused on methods of data transformation. RUL is predicted using Box-cox transforms and Monte Carlo simulations as described in Zhang Yongzhi et al, "Lithiumion battery remaining useful life prediction with Box-cox transformation and monte carlo simulation, IEEE Transactions on Industrial Electronics,66 (2): 1585-1597,2018. Some studies have been predicted by modeling lithium ion batteries. Dong Guangzhong et al, "Battery health prognosis using brownian motion modeling and particle filtering IEEE Transactions on Industrial Electronics,65 (11): 8646-8655,2018," use Brownian motion models and particle filters to predict SOH and RUL online. Guha Arijit and Patra Amit are described in "State of health estimation of lithium-ion batteries using capacity fade and internal resistance growth models IEEE Transactions on Transportation Electrification,4 (1): 135-146,2017," which utilize the characteristics of battery capacity fade and internal resistance to improve the performance of particle filters. Ma Yan et al in "Yan Ma, yang Chen, xiuwen Zhou, and Hong Chen. Remaining useful life prediction of lithium-ion battery based on Gauss-Hermite partial filter IEEE Transactions on Control Systems Technology,27 (4): 1788-1795,2018." use extended Kalman filtering to construct a predictive model, and use Gauss-Hermite particle filters to update the parameters of the Kalman filtering. Other studies have used machine learning methods. Wei Jingwen et al, in "Remaining useful life prediction and state of health diagnosis for lithium-ion batteries using particle filter and support vector regressions. IEEE Transactions on Industrial Electronics,65 (7): 5634-5643,2017," use a support vector machine and a particle filter to predict battery life. Khumprom Phattara and Yodo Nita use deep neural networks in "Adata-driven predictive prognostic model for lithium-ion batteries based on a deep learning algorithm.energy, 12 (4): 660,2019." to predict battery life. Severson et al, in "Data-driven prediction of battery cycle life before capacity degradation. Nature Energy,4 (5): 383-391,2019," describe the use of feature engineering and a linear neural network to solve this problem.
However, in the above prior art, when training a model, it is necessary to extract features from data and then train the model using the features, so the complexity of prediction is high, and the accuracy of prediction is low.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a lithium ion battery life prediction method based on stream data, which can reduce the complexity of prediction and improve the accuracy of prediction.
In order to achieve the above purpose, the method for predicting the life of the lithium ion battery based on stream data comprises the following steps:
acquiring stream data of battery characteristics, simultaneously constructing a plurality of random forest models, and training the random forest models by utilizing the stream data of the battery characteristics;
and acquiring a plurality of time series data of battery characteristics to be predicted, wherein one time series data corresponds to one random forest model, then inputting each time series data into the corresponding random forest model for life prediction, averaging life prediction results output by each random forest model, and finally taking the averaged results as final battery life prediction results.
Random forest model RF i Consisting of several different decision trees, a random forest model RF i Is the stream data of the i-th feature of the battery;
for a single decision tree, each node except for a leaf node is provided with an SVM classifier, data of t' time nodes are randomly extracted as input data of the SVM classifier on the basis of input data of the size of Q x t of the decision tree, the SVM classifier divides a battery into two types, each node and so on until the battery is divided into the leaf nodes, and at the moment, the leaf node is provided with a label which is the average value of the service lives of the batteries in the leaf node.
The training process of the SVM classifier is as follows:
1) A penalty parameter c=1 is chosen, a kernel function K (x i ,x j ) Constructing and solving convex quadratic regression problem for rbf functions.t./>Obtaining the optimal solution
2) Select alpha * Is a component of (a)Wherein (1)>Calculate->
3) The final classification decision function is obtained as follows:
characteristics of the battery include the voltage and temperature of the battery.
The invention has the following beneficial effects:
according to the lithium ion battery life prediction method based on the flow data, when the method is specifically operated, time series data such as voltage and temperature of a battery are directly used as input of a random forest model to train the random forest model, instead of the traditional training based on the characteristics extracted from the data, because characteristic engineering is not needed, the method is easy to implement, simple measurement is needed, complex data post-processing is not needed, the method can be used as input of an SVM classifier, the number of parameters to be adjusted is small, and compared with a method of extracting the characteristics first and then performing machine learning, parameter adjustment is easier, and prediction is more accurate.
Drawings
FIG. 1 is a schematic diagram of RUL predictive algorithm;
FIG. 2 is a block diagram of a random forest;
FIG. 3 is a schematic diagram of a decision tree training and prediction process;
fig. 4 is a schematic diagram of a tag dynamic update process.
Detailed Description
The invention is described in further detail below with reference to the attached drawing figures:
the algorithm for predicting the service life of the lithium ion battery comprises the following steps:
constructing a plurality of random forest models, wherein each random forest model carries out life prediction according to one input time series data (such as working condition data of voltage), then averages life prediction results output by each random forest model to obtain a final battery life prediction result, meanwhile, a decision tree in the random forest model utilizes a support vector machine to process splitting of decision tree nodes so as to ensure that even if the time series data are longer, the depth of the decision tree can be controlled, the specific principle is as shown in figure 1, q-axis represents different battery data, f-axis represents different battery characteristics, the battery characteristics are voltages and temperatures of different times of circulation, t-axis is a time axis, the invention uses stream data, and the data of each characteristic dimension are respectively sent into different random forest models (RF) 1 ，RF 2 ……RF f ) Training is carried out, and then the results are summarized.
Random forest model RF i Consisting of several different decision trees, a random forest model RF i The input of the decision tree is the stream data of the ith feature, the size is Q t, the input of each decision tree is obtained by a mode of sampling with a put back, the size is Q t, and the structure of the random forest is shown in fig. 2.
For a single decision tree, each node except for a leaf node is provided with an SVM classifier, based on the input data of the size of Q x t of the decision tree, the data of t' time nodes are randomly extracted as the input data of the SVM classifier, the SVM classifier divides batteries into two classes, each node and so on until the battery is divided into the leaf nodes, at the moment, the leaf nodes are provided with a label, the label is the average value of the service lives of the batteries in the leaf nodes, and when the model is applied, the label is used as the service life prediction value of the decision tree, the training and prediction process of the decision tree is shown in fig. 3, and the training process of the SVM classifier is as follows:
1) A penalty parameter c=1 is chosen, a kernel function K (x i ,x j ) Constructing and solving convex quadratic regression problem for rbf functions.t./>Obtaining the optimal solution
2) Select alpha * Is a component of (a)Wherein (1)>Calculate->
3) The final classification decision function is obtained as follows:
as an algorithm with supervised learning, the SVM classifier needs to have a-1/1 label for training data when the SVM classifier is trained, and the existing training data label is a battery life, so that the numerical label of the battery life needs to be dynamically converted into a classification label. For the cells contained in the input data of the Q SVM classifiers, the median of all the cell lives is found first, and for cells below the median, the remaining cells are marked-1, and the process is re-marked before each training of the SVM classifier, as shown in fig. 4.
Compared with the traditional algorithm for training the model based on the battery characteristics, the method directly uses the time series data such as the voltage and the temperature of the battery as the input of the model to train, is easier to implement because no characteristic engineering is needed, can be used as the input of the SVM classifier only by simple measurement and no complex data post-processing, and has fewer parameters to be adjusted, so that the method is easier to adjust compared with a method of extracting the characteristics and then performing machine learning.
Meanwhile, the traditional battery feature algorithm-based feature selection results in data requiring a plurality of cycles, which has a great influence on the model training efficiency. Meanwhile, the service life of the finished lithium ion battery is reduced due to excessive cycles in the test, however, the method has lower requirement on the number of times of data cycle, and is beneficial to quick and effective prediction in practical application.
At present, the invention has obtained excellent effects in a lithium ion battery working condition database established by Kristen A.Severson and Peter M.Attia et al, and specific indexes are as follows:
for the problem of predicting lithium ion battery life, the average absolute percentage error (MAPE) of the training set in the invention is 18.00%, and the MAPE of the test set is 15.16%. In contrast, the training set MAPE of the prediction model based on the LSTM algorithm in the same data set is 32.44%, and the test set MAPE is 24.87%; the training set MAPE of the prediction model based on the CNN algorithm is 18.56%, and the MAPE of the test set is 20.13%, so that the prediction accuracy is higher.
Claims (1)
1. The lithium ion battery life prediction method based on the stream data is characterized by comprising the following steps of:
acquiring stream data of battery characteristics, simultaneously constructing a plurality of random forest models, and training the random forest models by utilizing the stream data of the battery characteristics;
acquiring a plurality of time series data of battery characteristics to be predicted, wherein one time series data corresponds to one random forest model, then inputting each time series data into the corresponding random forest model for life prediction, averaging life prediction results output by each random forest model, and finally taking the averaged results as final battery life prediction results;
random forest model RF i Consisting of several different decision trees, a random forest model RF i The input of (2) isStream data of the i-th feature of the battery;
for a single decision tree, each node except for a leaf node is provided with an SVM classifier, based on the input data of the size of Q x t of the decision tree, the data of t' time nodes are randomly extracted to be used as the input data of the SVM classifier, the SVM classifier divides the battery into two types, each node and so on until the battery is divided into the leaf nodes, at the moment, the leaf node is provided with a label, and the label is the average value of the service life of the battery in the leaf node;
the training process of the SVM classifier is as follows:
1) A penalty parameter c=1 is chosen, a kernel function K (x i ,x j ) Constructing and solving convex quadratic regression problem for rbf functions.t./>Obtaining the optimal solution
2) Select alpha * Is a component of (a)Wherein (1)>Calculate->
3) The final classification decision function is obtained as follows:
characteristics of the battery include the voltage and temperature of the battery.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010797577.5A CN112100902B (en) | 2020-08-10 | 2020-08-10 | Lithium ion battery life prediction method based on flow data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010797577.5A CN112100902B (en) | 2020-08-10 | 2020-08-10 | Lithium ion battery life prediction method based on flow data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112100902A CN112100902A (en) | 2020-12-18 |
| CN112100902B true CN112100902B (en) | 2023-12-22 |
Family
ID=73753191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010797577.5A Active CN112100902B (en) | 2020-08-10 | 2020-08-10 | Lithium ion battery life prediction method based on flow data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112100902B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113589175B (en) * | 2021-08-23 | 2024-04-12 | 上海智能新能源汽车科创功能平台有限公司 | Fuel cell life prediction method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846227A (en) * | 2017-12-05 | 2018-11-20 | 北京航空航天大学 | A kind of capacity of lithium ion battery degradation prediction appraisal procedure based on random forest and capacity self- recoverage effect analysis |
| CN110888059A (en) * | 2019-12-03 | 2020-03-17 | 西安科技大学 | Algorithm based on improved random forest combined cubature Kalman power battery state of charge estimation |
| AU2020100709A4 (en) * | 2020-05-05 | 2020-06-11 | Bao, Yuhang Mr | A method of prediction model based on random forest algorithm |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019017991A1 (en) * | 2017-07-21 | 2019-01-24 | Quantumscape Corporation | Predictive model for estimating battery states |
-
2020
- 2020-08-10 CN CN202010797577.5A patent/CN112100902B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846227A (en) * | 2017-12-05 | 2018-11-20 | 北京航空航天大学 | A kind of capacity of lithium ion battery degradation prediction appraisal procedure based on random forest and capacity self- recoverage effect analysis |
| CN110888059A (en) * | 2019-12-03 | 2020-03-17 | 西安科技大学 | Algorithm based on improved random forest combined cubature Kalman power battery state of charge estimation |
| AU2020100709A4 (en) * | 2020-05-05 | 2020-06-11 | Bao, Yuhang Mr | A method of prediction model based on random forest algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 韦振汉 ; 宋树祥 ; 夏海英 ; .基于随机森林的锂离子电池荷电状态估算.广西师范大学学报(自然科学版).2018,(04),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112100902A (en) | 2020-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Luo et al. | A review of deep learning approach to predicting the state of health and state of charge of lithium-ion batteries | |
| Khan et al. | Batteries state of health estimation via efficient neural networks with multiple channel charging profiles | |
| CN110187290B (en) | Lithium ion battery residual life prediction method based on fusion algorithm | |
| Liu et al. | Transfer learning for battery smarter state estimation and ageing prognostics: Recent progress, challenges, and prospects | |
| Zhou et al. | Battery health prognosis using improved temporal convolutional network modeling | |
| CN112508442B (en) | Transient stability assessment method and system based on automatic and interpretable machine learning | |
| Hu et al. | Electrochemical-theory-guided modeling of the conditional generative adversarial network for battery calendar aging forecast | |
| CN109740859A (en) | Transformer condition evaluation and system based on Principal Component Analysis and support vector machines | |
| WO2021017416A1 (en) | Deep compression power lithium battery fault diagnosis method under perceptual adversarial generation | |
| CN111707956B (en) | Method for predicting health state and residual life of multi-type lithium ion battery pack management system | |
| Wu et al. | Optimized multi-source fusion based state of health estimation for lithium-ion battery in fast charge applications | |
| Liu et al. | Telecommunication traffic prediction based on improved LSSVM | |
| Li et al. | An optimal stacking ensemble for remaining useful life estimation of systems under multi-operating conditions | |
| Zou et al. | Advancements in Artificial Neural Networks for health management of energy storage lithium-ion batteries: A comprehensive review | |
| CN115204038A (en) | Energy storage lithium battery life prediction method based on data decomposition and integration model | |
| Li et al. | Structure optimization for echo state network based on contribution | |
| CN112100902B (en) | Lithium ion battery life prediction method based on flow data | |
| Shen et al. | Transfer learning-based state of charge and state of health estimation for Li-ion batteries: A review | |
| Xia et al. | Historical data-independent remaining useful life prediction method based on dual-input deep learning neural network | |
| Wang et al. | A novel capacity estimation method for li-ion battery cell by applying ensemble learning to extremely sparse significant points | |
| CN117192417A (en) | XGBoost-based lithium ion battery health state prediction method | |
| Eleftheriadis et al. | Transfer Learning Techniques for the Lithium-Ion Battery State of Charge Estimation | |
| Li et al. | Degradation Pattern Recognition and Features Extrapolation for Battery Capacity Trajectory Prediction | |
| Zhou et al. | Combined sparse Bayesian learning strategy for remaining useful life forecasting of lithium-ion battery | |
| CN114219118A (en) | Method and system for predicting service life of intelligent electric meter based on D-S evidence theory |
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 |