Lithium battery health state evaluation system
By using a decision tree model in the lithium battery health status evaluation system and combining the various parameters of the lithium battery for evaluation, the problem that the existing technology cannot accurately predict the remaining service life and working ability of the lithium battery is solved, and higher evaluation accuracy is achieved and adaptability reduces the safety risks in the process of cascade utilization of lithium batteries.
Patent Information
- Application Number
- CN202411209433.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2025-02-14
AI Technical Summary
The existing lithium battery health status evaluation method cannot accurately predict the remaining service life and working capacity of lithium batteries, resulting in the failure to fully utilize the potential of lithium batteries during the cascade utilization process or increase safety risks, and the evaluation technology updates and iteration speed is slow, making it difficult to adapt. The rapid development of battery technology and application scenarios.
Using the decision tree model training unit, the decision tree model verification unit and the decision tree model application unit, data preprocessing, feature extraction and decision tree model construction are carried out by monitoring the voltage, current, temperature and charge and discharge times of lithium batteries, and data preprocessing, feature extraction and decision tree model construction are improved to improve the evaluation system accuracy and adaptability.
Through the application of the decision tree model, the health status of lithium batteries can be more accurately evaluated, the accuracy and adaptability of the evaluation system can be improved, so that it can keep up with the update and iteration speed of evaluation technology and methods, and reduce the safety of lithium batteries in the process of cascade utilization of lithium batteries risk.
Smart Images
Figure CN119438901A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a lithium battery health status assessment system. Background Art
[0002] The cascade utilization of retired lithium batteries is an urgent technical problem to be solved. From the application level, the most important thing to solve is the adaptability of lithium battery status and application scenarios. By selecting appropriate application scenarios for lithium batteries in different states, the stress in the cascade utilization of retired lithium batteries can be reduced, the service life of retired lithium batteries can be extended, and the safety risks in the cascade utilization process can be reduced. An important prerequisite for this work is to fully understand the remaining use value and working capacity of the battery through the health status assessment of the lithium battery.
[0003] Existing assessment methods may not accurately predict the remaining service life and working capacity of lithium batteries, resulting in the failure to fully utilize the potential of batteries during cascade utilization, or excessive use increasing safety risks. Existing assessment technologies and methods are updated and iterated slowly, making it difficult to quickly adapt to the rapid development of battery technology and application scenarios.
[0004] Therefore, there is a need for a lithium battery health status assessment system with autonomous training capabilities, which can calculate the RUL value based on multiple parameters of the lithium battery, increase the accuracy and adaptability of the assessment system, and enable it to keep up with the speed of updating and iteration of assessment technologies and methods. Summary of the invention
[0005] The purpose of the present invention is to solve the deficiencies of the above prior art and provide a lithium battery health status assessment system. A decision tree model is established through a decision tree model training unit, a decision tree model verification unit and a decision tree model application unit to increase the accuracy and adaptability of the assessment system so that it can keep up with the speed of updating and iterating the assessment technology and method. The following is a specific solution:
[0006] A lithium battery health status assessment system comprises a monitoring unit, a data preprocessing unit, a feature extraction unit, a decision tree model training unit, a decision tree model verification unit, a decision tree model application unit and a health status output unit, wherein the monitoring unit is used to collect the voltage, current, temperature and charge and discharge times of the lithium battery; the data preprocessing unit is used to clean, standardize and normalize the data of the monitoring unit; the feature extraction unit is used to extract the data processed by the data processing unit and extract features; the decision tree model training unit is used to use the features extracted by the feature extraction unit and construct a decision tree model; the decision tree model verification unit is used to verify the decision tree model by cross-validation or bootstrapping; the decision tree model application unit is used to apply the verified decision tree model to new lithium battery status data to predict its health status; and the health status output unit is used to output the health status of the lithium battery to a user according to the prediction result of the decision tree model.
[0007] Furthermore, the normalization is calculated based on the following formula:
[0008] X'=\frac{X-X_{min}}{X_{max}-X_{min}}, where (X') is the normalized data, X is the original data, X_{min} and X_{max} are the minimum and maximum values in the data set respectively.
[0009] Furthermore, the feature extraction unit quantifies the importance of the feature based on the following formula:
[0010] g_k=\frac{IG_k}{IV_k}, where IG_k is the information gain of feature k and IV_k is the intrinsic variance of feature k.
[0011] Furthermore, the specific implementation method of the decision tree model training unit is as follows: S1: removing features that have been used for data segmentation from the training data;
[0012] S2: Calculate the information gain ratio of the remaining features and select the feature with the highest information gain ratio;
[0013] S3: Use the selected features to separate the data and generate subtrees;
[0014] S4: Calculate the potential (T) on each subtree based on the following formula:
[0015] [T(\alpha)=\alpha|D|+\sum_{j=1}^{n}p_jE_j], where (\alpha) is the complexity parameter, D is the dataset in the node, p_j is the frequency of the jth class in the entire dataset, and E_j is the purity loss required to assign D_j to the jth class;
[0016] S5: Adjust the value of \alpha to obtain the subtree with the minimum cost;
[0017] S6: Remove the subtrees that increase the total cost and merge the remaining subtrees into a decision tree.
[0018] Furthermore, the decision tree model verification unit verifies the decision tree model based on the following formula:
[0019] Acc = \frac{TP+TN}{TP+TN+FP+FN}, where TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, and FN is the number of false negative examples. By calculating the accuracy rate ACC, we can measure the proportion of samples correctly classified by the model to the total samples.
[0020] Prec=\frac{TP}{TP+FP, the precision Prec is calculated to measure the proportion of positive examples predicted by the model to be positive examples, which reflects the prediction quality of the model;
[0021] Rec=\frac{TP}{TP+FN}. The recall rate Rec is calculated to measure the proportion of samples that are actually positive examples that are correctly predicted by the model as positive examples, reflecting the monitoring ability of the model.
[0022] Furthermore, the decision tree model application unit is calculated based on the following formula:
[0023] [\hat{y}=3.5+0.1\cdotV+0.01\cdotT-0.3\cdotC], where V is the voltage of the lithium battery, T is the temperature, and C is the number of charge and discharge cycles.
[0024] Furthermore, the health status output unit is connected to the decision tree model application unit, and the RUL value is calculated through the decision tree model. When RUL≥80%, the health status output unit outputs healthy; when 50%≤RUL<80%, the health status output unit outputs decay; when RUL<50%, the health status output unit outputs fault.
[0025] Beneficial effects: The monitoring unit comprehensively collects key parameters of lithium batteries such as voltage, current, temperature, and charge and discharge times, providing sufficient data support for health status assessment;
[0026] The data preprocessing unit cleans, normalizes and normalizes the original data to improve data quality and provide accurate and reliable input for feature extraction and model training;
[0027] The feature extraction unit extracts key features reflecting the battery health status from the processed data to improve the accuracy and efficiency of the assessment;
[0028] The decision tree model training unit, decision tree model verification unit and decision tree model application unit use the feature selection method to construct the decision tree model, which improves the accuracy and generalization ability of the model, increases the precision and adaptability of the evaluation system, and has autonomous training capabilities, so that it can keep up with the speed of updating and iteration of evaluation technologies and methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] Figure 1 It is a schematic diagram of the structure of the lithium battery health status assessment system;
[0030] Figure 2 It is a flow chart of the decision tree model training unit of the lithium battery health status assessment system. DETAILED DESCRIPTION
[0031] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with embodiments and drawings. The embodiments are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.
[0032] In this embodiment:
[0033] Please see Figure 1-Figure 2 A lithium battery health status assessment system includes a monitoring unit, a data preprocessing unit, a feature extraction unit, a decision tree model training unit, a decision tree model verification unit, a decision tree model application unit and a health status output unit, wherein the monitoring unit is used to collect the voltage, current, temperature and charge and discharge times of the lithium battery; the data preprocessing unit is used to clean, standardize and normalize the data of the monitoring unit; the feature extraction unit is used to extract the data processed by the data processing unit and extract features; the decision tree model training unit is used to use the features extracted by the feature extraction unit and construct a decision tree model; the decision tree model verification unit is used to verify the decision tree model through cross-validation or bootstrap method; the decision tree model application unit is used to apply the verified decision tree model to new lithium battery status data to predict its health status; the health status output unit is used to output the health status of the lithium battery to the user according to the prediction result of the decision tree model.
[0034] Normalization is calculated based on the following formula:
[0035] X'=\frac{X-X_{min}}{X_{max}-X_{min}}, where (X') is the normalized data, X is the original data, X_{min} and X_{max} are the minimum and maximum values in the data set respectively.
[0036] The feature extraction unit quantifies the importance of features based on the following formula:
[0037] g_k=\frac{IG_k}{IV_k}, where IG_k is the information gain of feature k and IV_k is the intrinsic variance of feature k.
[0038] The specific implementation method of the decision tree model training unit is as follows: S1: remove the features that have been used for data segmentation from the training data;
[0039] S2: Calculate the information gain ratio of the remaining features and select the feature with the highest information gain ratio;
[0040] S3: Use the selected features to separate the data and generate subtrees;
[0041] S4: Calculate the potential (T) on each subtree based on the following formula:
[0042] [T(\alpha)=\alpha|D|+\sum_{j=1}^{n}p_jE_j], where (\alpha) is the complexity parameter, D is the dataset in the node, p_j is the frequency of the jth class in the entire dataset, and E_j is the purity loss required to assign D_j to the jth class;
[0043] S5: Adjust the value of \alpha to obtain the subtree with the minimum cost;
[0044] S6: Remove the subtrees that increase the total cost and merge the remaining subtrees into a decision tree.
[0045] The decision tree model validation unit validates the decision tree model based on the following formula:
[0046] Acc = \frac{TP+TN}{TP+TN+FP+FN}, where TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, and FN is the number of false negative examples. By calculating the accuracy rate ACC, we can measure the proportion of samples correctly classified by the model to the total samples.
[0047] Prec=\frac{TP}{TP+FP, the precision Prec is calculated to measure the proportion of positive examples predicted by the model to be positive examples, which reflects the prediction quality of the model;
[0048] Rec=\frac{TP}{TP+FN}. The recall rate Rec is calculated to measure the proportion of samples that are actually positive examples that are correctly predicted by the model as positive examples, reflecting the monitoring ability of the model.
[0049] The decision tree model application unit is calculated based on the following formula:
[0050] [\hat{y}=3.5+0.1\cdotV+0.01\cdotT-0.3\cdotC], where V is the voltage of the lithium battery, T is the temperature, and C is the number of charge and discharge cycles.
[0051] The health status output unit is connected to the decision tree model application unit, and the RUL value is calculated by the decision tree model. When RUL≥80%, the health status output unit outputs health; when 50%≤RUL<80%, the health status output unit outputs decay; when RUL<50%, the health status output unit outputs failure.
[0052] The following is a specific implementation method:
[0053] (1) Data collection: The monitoring unit collects key parameters of the lithium battery in real time, such as voltage, current, temperature, and charge and discharge times;
[0054] (2) Data preprocessing: Clean, normalize and normalize the collected raw data to improve the quality and consistency of the data;
[0055] (3) Feature extraction: extract key features from the preprocessed data and quantify the importance of the features based on information gain and intrinsic variance;
[0056] (4) Model training: Use the extracted features to build a decision tree model and optimize the model structure by removing features used for data segmentation, selecting features with the highest information gain ratio, and calculating subtree potential.
[0057] (5) Model validation: Use methods such as cross-validation or bootstrapping to verify the correctness of the decision tree model and evaluate performance indicators such as accuracy, precision, and recall;
[0058] (6) State prediction: Apply the verified decision tree model to new battery state data to predict the battery health state;
[0059] (7) Health status output: Calculate the remaining useful life (RUL) of the battery based on the model prediction results.
[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.
Claims
1. A lithium battery health status assessment system, characterized in that: It includes a monitoring unit, a data preprocessing unit, a feature extraction unit, a decision tree model training unit, a decision tree model verification unit, a decision tree model application unit and a health status output unit. The monitoring unit is used to collect the voltage, current, temperature and charge and discharge times of the lithium battery; the data preprocessing unit is used to clean, standardize and normalize the data of the monitoring unit; the feature extraction unit is used to extract the data processed by the data processing unit and extract features; the decision tree model training unit is used to use the features extracted by the feature extraction unit and construct a decision tree model; the decision tree model verification unit is used to verify the decision tree model by cross-validation or bootstrap method; the decision tree model application unit is used to apply the verified decision tree model to new lithium battery status data to predict its health status; the health status output unit is used to output the health status of the lithium battery to the user according to the prediction result of the decision tree model.
2. A lithium battery health status assessment system according to claim 1, characterized in that , the normalization is calculated based on the following formula: X'=\frac{X-X_{min}}{X_{max}-X_{min}}, where (X') is the normalized data, X is the original data, X_{min} and X_{max} are the minimum and maximum values in the data set respectively.
3. A lithium battery health status assessment system according to claim 1, characterized in that: The feature extraction unit quantifies the importance of features based on the following formula: g_k=\frac{IG_k}{IV_k}, where IG_k is the information gain of feature k and IV_k is the intrinsic variance of feature k.
4. A lithium battery health status assessment system according to claim 1, characterized in that: The specific implementation method of the decision tree model training unit is as follows: S1: removing features that have been used for data segmentation from the training data; S2: Calculate the information gain ratio of the remaining features and select the feature with the highest information gain ratio; S3: Use the selected features to separate the data and generate subtrees; S4: Calculate the potential (T) on each subtree based on the following formula: [T(\alpha)=\alpha|D|+\sum_{j=1}^{n}p_jE_j], where (\alpha) is the complexity parameter, D is the dataset in the node, p_j is the frequency of the jth class in the entire dataset, and E_j is the purity loss required to assign D_j to the jth class; S5: Adjust the value of \alpha to obtain the subtree with the minimum cost; S6: Remove the subtrees that increase the total cost and merge the remaining subtrees into a decision tree.
5. A lithium battery health status assessment system according to claim 1, characterized in that: The decision tree model verification unit verifies the decision tree model based on the following formula: Acc = \frac{TP+TN}{TP+TN+FP+FN}, where TP is the number of true positive examples, TN is the number of true negative examples, FP is the number of false positive examples, and FN is the number of false negative examples. By calculating the accuracy rate ACC, we can measure the proportion of samples correctly classified by the model to the total samples. Prec=\frac{TP}{TP+FP, the precision Prec is calculated to measure the proportion of positive examples predicted by the model to be positive examples, which reflects the prediction quality of the model; Rec=\frac{TP}{TP+FN}. The recall rate Rec is calculated to measure the proportion of samples that are actually positive examples that are correctly predicted by the model as positive examples, reflecting the monitoring ability of the model.
6. A lithium battery health status assessment system according to claim 1, characterized in that: The decision tree model application unit is calculated based on the following formula: [\hat{y}=3.5+0.1\cdotV+0.01\cdotT-0.3\cdotC], where V is the voltage of the lithium battery, T is the temperature, and C is the number of charge and discharge cycles.
7. A lithium battery health status assessment system according to claim 1, characterized in that: The health status output unit is connected to the decision tree model application unit, and the RUL value is calculated by the decision tree model. When RUL≥80%, the health status output unit outputs health; when 50%≤RUL<80%, the health status output unit outputs decline; when RUL<50%, the health status output unit outputs failure.