CN114429182A - Retired power battery grade classification method based on improved CART algorithm - Google Patents
Retired power battery grade classification method based on improved CART algorithm Download PDFInfo
- Publication number
- CN114429182A CN114429182A CN202210049882.5A CN202210049882A CN114429182A CN 114429182 A CN114429182 A CN 114429182A CN 202210049882 A CN202210049882 A CN 202210049882A CN 114429182 A CN114429182 A CN 114429182A
- Authority
- CN
- China
- Prior art keywords
- power battery
- retired power
- decision tree
- retired
- attribute
- 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
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000003066 decision tree Methods 0.000 claims abstract description 54
- 238000011156 evaluation Methods 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 abstract description 20
- 238000013138 pruning Methods 0.000 abstract description 6
- 230000011218 segmentation Effects 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000012535 impurity Substances 0.000 description 2
- 238000007599 discharging Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- General Engineering & Computer Science (AREA)
- Primary Health Care (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Water Supply & Treatment (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Secondary Cells (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
A retired power battery grade screening method based on an improved CART decision tree algorithm aims to apply the improved CART decision tree algorithm to retired power battery grade classification, improve grade classification efficiency and lay a certain foundation for retired power battery gradient utilization. The method comprises the following steps: firstly, when an attribute segmentation point is selected, determining an optimal threshold point of each characteristic attribute of a retired power battery by combining a Fayyad boundary point judgment theorem, and calculating a Gini coefficient at the optimal threshold point; then, taking the optimal threshold point of each characteristic attribute of the retired power battery as a splitting condition of the decision tree, sequentially performing binary splitting until the retired power batteries in all leaf nodes belong to the same class, and generating the decision tree; and finally, pruning the decision tree by adopting a cost complexity pruning algorithm, preventing overfitting, improving the generalization performance of the decision tree and obtaining the retired power battery grade screening optimal decision tree.
Description
Technical Field
The invention relates to the technical field of recycling of retired batteries, in particular to a class classification technology of retired power batteries of electric automobiles.
Background
By 2035 years, pure electric vehicles become the mainstream of new vehicles, the vehicles in the public field are fully electric, and the holding capacity of the electric vehicles is continuously increased greatly. However, after the power battery for the vehicle is used for several years, the performance of the power battery for the vehicle is gradually reduced, and when the normal use requirement of the electric vehicle cannot be met, the power battery for the vehicle can be retired from the electric vehicle. Under different use environments, the attenuation degree of the power battery is different, and if the power battery is directly utilized in a gradient mode without screening and classification, risks such as overcharge, overdischarge and even explosion of the battery can be caused. Under the background, how to realize the screening and classification of the retired batteries of the electric vehicles is a key problem which needs to be solved urgently.
The existing classification methods mainly comprise single-parameter screening, multi-parameter screening and characteristic curve screening. The single-parameter screening method only measures one parameter and often cannot represent the overall performance of the retired power battery. The multi-parameter screening method completes screening by measuring a plurality of parameters, has good screening effect, but has low screening efficiency. The characteristic curve screening method is to apply specific input to the retired power battery, and then compare the charging and discharging characteristic curves of the retired power battery to realize screening. The screening effect of the characteristic curve method is superior to that of the former two methods, but the characteristic curve is relatively difficult to obtain, so that the method cannot effectively screen large-scale retired power batteries.
According to the method, the CART decision tree algorithm and the Fayyad boundary point judgment theorem are combined, the algorithm calculation amount is reduced by selecting the attribute optimal threshold value point, and the improved CART decision tree algorithm is applied to the rank screening of the retired power battery, so that the screening efficiency is improved. Compared with a single-parameter screening method, a multi-parameter screening method and a characteristic curve screening method, the method can improve the efficiency of classification of the retired power battery grades under the condition of keeping higher accuracy.
Disclosure of Invention
The invention aims to apply the improved CART decision tree algorithm to the grade classification of the retired power battery, improve the grade classification efficiency and lay a certain foundation for the gradient utilization of the retired power battery.
The invention relates to a retired power battery grade classification method based on an improved CART algorithm, which comprises the following steps:
dividing a retired power battery sample set, using 60% of retired power battery sample data for model training, and using 40% of sample data for model performance evaluation;
step (2) selecting the capacity A1Internal resistance A2And secondary cycle life A3The method is used as a characteristic basis for classifying grades of retired power batteries of the electric automobile;
combining Fayyad boundary point judgment theorem, and when the attribute division point of the retired power battery is selected, regarding the continuous characteristic attribute A1、A2、A3Respectively enabling all samples of the decision tree nodes to be according to the attribute A in order to find a threshold point which enables the average class entropy of the retired power battery sample set to reach the minimum value1、A2、A3After the specific numerical values are arranged in an ascending order, the boundary point between two adjacent heterogeneous samples is the attribute A1、A2、A3The optimal threshold point of (a);
when the attribute division point of the retired power battery is selected in the step (4), the attribute division point is not selectedThe Gini coefficient at each division point needs to be calculated, and only the Gini coefficient at the characteristic attribute threshold point of the retired power battery needs to be calculated; taking the characteristic attribute threshold point with the minimum Gini coefficient as a root node of a decision tree, and dividing an initial retired power battery data set S into two data subsets S1And S2To S1And S2Continuing recursion, and establishing child nodes of a decision tree of retired power batteries until the retired power batteries in all the child nodes belong to the same class;
step 5, finding the maximum decision tree depth by investigating the relationship between the maximum decision tree depth and the classification accuracy of the retired power battery grade; the relation between the cost complexity parameter and the total purity of leaf nodes is inspected, the decision tree is pruned, and overfitting is prevented;
step (6) generating a grade classification decision tree according to the optimal decision tree parameters of the classification of the retired power battery, namely the maximum decision tree depth and the optimal cost complexity parameters, so as to realize the classification of the retired power battery;
and (7) according to the classification result of the improved CART decision tree algorithm, using the retired power battery in an application scene with lower performance requirements according to the capacity and the internal resistance of the retired power battery relative to the rated capacity and the rated internal resistance.
The invention has the advantages that: the invention provides a retired battery grade classification method based on an improved CART algorithm based on the characteristic attribute of a retired power battery. The traditional CART algorithm is combined with the Fayyad boundary point judgment theorem, and only Gini coefficients at the optimal segmentation threshold point are calculated when the internal segmentation points are selected, so that the problem of large calculation amount of the traditional CART algorithm is solved, and the efficiency of grade classification is improved.
Drawings
Fig. 1 is a flowchart of classification of retired battery levels based on the CART algorithm, and fig. 2 is a relationship between maximum depth and accuracy of a decision tree.
Detailed Description
The invention relates to a retired power battery grade classification method based on an improved CART algorithm, which comprises the following steps:
dividing a retired power battery sample set, using 60% of retired power battery sample data for model training, and using 40% of sample data for model performance evaluation;
step (2) selecting the capacity A1Internal resistance A2And secondary cycle life A3The method is used as a characteristic basis for classifying the grades of the retired power batteries of the electric vehicles;
combining Fayyad boundary point judgment theorem, and when the attribute division point of the retired power battery is selected, regarding the continuous characteristic attribute A1、A2、A3Respectively enabling all samples of the decision tree nodes to be according to the attribute A in order to find a threshold point which enables the average class entropy of the retired power battery sample set to reach the minimum value1、A2、A3After the specific numerical values are arranged in an ascending order, the boundary point between two adjacent heterogeneous samples is the attribute A1、A2、A3The optimal threshold point of (a);
when selecting the attribute division points of the retired power battery, only calculating the Gini coefficient at the characteristic attribute threshold point of the retired power battery without calculating the Gini coefficient at each division point; taking a characteristic attribute threshold point with the minimum Gini coefficient as a root node of a decision tree, and dividing an initial retired power battery data set S into two data subsets S1And S2To S1And S2Continuing recursion, and establishing child nodes of a decision tree of retired power batteries until the retired power batteries in all the child nodes belong to the same class;
step 5, finding the maximum decision tree depth by investigating the relationship between the maximum decision tree depth and the classification accuracy of the retired power battery grade; the relation between the cost complexity parameter and the total purity of leaf nodes is inspected, the decision tree is pruned, and overfitting is prevented;
step (6) generating a grade classification decision tree according to the optimal decision tree parameters of the classification of the retired power battery, namely the maximum decision tree depth and the optimal cost complexity parameters, so as to realize the classification of the retired power battery;
and (7) according to the classification result of the improved CART decision tree algorithm, using the retired power battery in an application scene with lower performance requirements according to the capacity and the internal resistance of the retired power battery relative to the rated capacity and the rated internal resistance.
The present invention will be described in further detail with reference to the accompanying drawings.
The invention discloses a retired power battery grade classification method based on an improved CART algorithm, which has the flow shown in figure 1 and specifically comprises the following steps:
selecting 200 ex-service power battery samples of 4 electric buses running for three years, wherein 60% of the ex-service power battery sample data is used for model training, and 40% of the sample data is used for model performance evaluation;
selecting the capacity, the internal resistance and the secondary cycle life as characteristic bases for classifying grades of retired power batteries of the electric automobile;
step (3) when the optimal partition threshold value of the properties of the retired power battery is selected, combining a Fayyad boundary point judgment theorem; for the continuous attribute A, the T which enables the average class entropy of the sample set to reach the minimum value is always positioned between two adjacent heterogeneous samples in the sorted sample sequence, namely the T which enables the average class entropy of the sample set to reach the minimum value is a dividing point of the attribute A; for a boundary point T of the characteristic attribute A of the retired power battery, the average class entropy for dividing the retired power battery sample S is defined as follows:
wherein S isaA subset with the value of less than or equal to T on the attribute A is taken as a retired power battery sample set SbA subset greater than T. Ent is information entropy;
the optimal splitting point of the characteristic attribute of the retired power battery is positioned at the boundary point of adjacent heterogeneous batteries; in order to obtain the optimal splitting point, the average class entropy for dividing the sample set is required to be minimum; the impurity degree of the sample is expressed by entropy, and the larger the entropy value is, the higher the impurity degree of the subset division is. The information entropy is defined as follows:
wherein, PiRepresents the i-th type sample CiThe proportion of S;
in the step (4), under the same binary splitting condition, the entropy and the Gini coefficient have basically the same change trend, and the smaller the entropy is, the smaller the Gini coefficient is; therefore, Gini coefficients at all the segmentation points of the characteristic attributes of the retired power battery do not need to be calculated, and Gini coefficients at different boundary points of the retired power battery only need to be calculated; selecting the attribute with the minimum Gini coefficient as a decision tree root node; then dividing the initial battery data set S into two battery data subsets S according to the characteristic attributes of the batteries in the retired power battery data set1And S2To S1And S2Continuing recursion, and establishing child nodes of the retired power battery classification decision tree until the retired power batteries in all the child nodes belong to the same class;
step 5, finding the maximum decision tree depth by investigating the relation between the maximum decision tree depth and the accuracy; the relation between the decision tree depth and the battery classification prediction accuracy is shown in fig. 2, and when the decision tree depth is 4, the classification prediction accuracy of the retired battery reaches the maximum; with the increase of the depth of the decision tree, the prediction accuracy rate does not change obviously any more, and the maximum depth of the decision tree is determined to be 4;
step (6) pruning the decision tree by adopting a cost complexity pruning algorithm; according to the cost complexity pruning algorithm, a sub-tree sequence is formed through pruning, then an independent retired power battery is utilized to verify a data set, the prediction error of each sub-tree in the sub-tree sequence is tested, and the decision tree with the minimum prediction error is used as the optimal decision tree;
step (7) generating an optimal decision tree suitable for the class classification of the retired power battery according to the obtained optimal decision tree parameters of the class classification of the retired power battery; obtaining a method capable of improving the classification efficiency of retired power battery grades under the condition of keeping higher accuracy;
the above is one of the implementation methods of the present invention, and it is obvious to a person skilled in the art that various changes can be made to the above embodiments without any creative effort, and the object of the present invention can be achieved. It will be apparent that such variations are intended to be included within the scope of the invention as defined in the claims.
Claims (1)
1. The retired power battery grade classification method based on the improved CART algorithm is characterized by comprising the following steps:
dividing a retired power battery sample set, wherein 60% of retired power battery sample data is used for model training, and 40% of sample data is used for model performance evaluation;
step (2) selecting the capacity A1Internal resistance A2And secondary cycle life A3The method is used as a characteristic basis for classifying grades of retired power batteries of the electric automobile;
combining Fayyad boundary point judgment theorem, and when the attribute division point of the retired power battery is selected, regarding the continuous characteristic attribute A1、A2、A3Respectively enabling all samples of the decision tree nodes to be according to the attribute A in order to find a threshold point which enables the average class entropy of the retired power battery sample set to reach the minimum value1、A2、A3After the specific numerical values are arranged in an ascending order, the boundary point between two adjacent heterogeneous samples is the attribute A1、A2、A3The optimal threshold point of (a);
when selecting the attribute division points of the retired power battery, only calculating the Gini coefficient at the characteristic attribute threshold point of the retired power battery without calculating the Gini coefficient at each division point; taking the characteristic attribute threshold point with the minimum Gini coefficient as a root node of a decision tree, and dividing an initial retired power battery data set S into two data subsets S1And S2To S1And S2Continuing recursion, and establishing child nodes of a decision tree of retired power batteries until the retired power batteries in all the child nodes belong to the same class;
step 5, finding the maximum decision tree depth by investigating the relationship between the maximum decision tree depth and the classification accuracy of the retired power battery grade; the relation between the cost complexity parameter and the total purity of leaf nodes is inspected, the decision tree is pruned, and overfitting is prevented;
step (6) generating a grade classification decision tree according to the optimal decision tree parameters of the classification of the retired power battery, namely the maximum decision tree depth and the optimal cost complexity parameters, so as to realize the classification of the retired power battery;
and (7) according to the classification result of the improved CART decision tree algorithm, using the retired power battery in an application scene with lower performance requirements according to the capacity and the internal resistance of the retired power battery relative to the rated capacity and the rated internal resistance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210049882.5A CN114429182A (en) | 2022-01-17 | 2022-01-17 | Retired power battery grade classification method based on improved CART algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210049882.5A CN114429182A (en) | 2022-01-17 | 2022-01-17 | Retired power battery grade classification method based on improved CART algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114429182A true CN114429182A (en) | 2022-05-03 |
Family
ID=81312946
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210049882.5A Pending CN114429182A (en) | 2022-01-17 | 2022-01-17 | Retired power battery grade classification method based on improved CART algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114429182A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115877230A (en) * | 2022-11-30 | 2023-03-31 | 上海玫克生储能科技有限公司 | Method, system, device and medium for determining fault of battery module |
-
2022
- 2022-01-17 CN CN202210049882.5A patent/CN114429182A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115877230A (en) * | 2022-11-30 | 2023-03-31 | 上海玫克生储能科技有限公司 | Method, system, device and medium for determining fault of battery module |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112103580B (en) | Lithium battery charging method based on equivalent internal resistance | |
| CN114114039B (en) | Method and device for evaluating consistency of single battery cells of battery system | |
| CN113887601B (en) | Retired power battery recombination method based on cluster sorting | |
| CN107741568B (en) | Lithium battery SOC estimation method based on state transition optimization RBF neural network | |
| CN109711483B (en) | Spark Autoencoder-based power system operation mode clustering method | |
| CN113715629B (en) | Residual driving range prediction method based on improved symbolic regression and XGBoost algorithm | |
| CN103699678B (en) | A kind of hierarchy clustering method based on multistage stratified sampling and system | |
| CN116842459B (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
| CN116449218B (en) | Lithium battery health state estimation method | |
| CN112287980B (en) | Power battery screening method based on typical feature vector | |
| CN112791997A (en) | Method for cascade utilization and screening of retired battery | |
| CN110490263A (en) | Battery sorting method based on charging and discharging curve and fuzzy clustering | |
| CN113033596A (en) | Refined identification method for user electricity consumption behavior category and typical electricity consumption mode | |
| CN114429182A (en) | Retired power battery grade classification method based on improved CART algorithm | |
| CN107895212A (en) | Lead-acid battery life-span prediction method based on sliding window and various visual angles Fusion Features | |
| CN114332825B (en) | Road terrain distribution identification method and device based on deep learning and storage medium | |
| CN112305441A (en) | Power battery health state assessment method under integrated clustering | |
| CN114487835B (en) | Retired power battery grading method based on particle swarm optimization BIRCH algorithm | |
| CN115688006A (en) | Retired battery screening and classifying method | |
| CN114355218A (en) | Lithium ion battery charge state prediction method based on multi-feature quantity screening | |
| CN111693868A (en) | Lithium battery state of charge estimation method based on density feature clustering integration | |
| CN117609818A (en) | Power grid association relation discovery method based on clustering and information entropy | |
| CN115598531A (en) | Method for estimating state of health of vehicle battery and training method of estimation model thereof | |
| CN114185956A (en) | Data mining method based on canty and k-means algorithm | |
| CN113537366A (en) | Transient stability evaluation method for power system |
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 |