Lithium battery health state classification method based on learning
By collecting lithium battery data and calculating uncertainty, a classification strategy was developed, which solved the problem that non-professional users had difficulty assessing the health status of lithium batteries. This enabled a simple and accurate assessment of the health status of lithium batteries, thus increasing its accessibility.
Patent Information
- Application Number
- CN202511420844.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-30
- Publication Date
- 2025-12-12
AI Technical Summary
Existing technologies cannot provide accurate assessments of the health status of lithium batteries for non-professional users through simple means, resulting in low adoption rates.
By collecting lithium battery data, calculating uncertainty, and formulating classification strategies, non-professional users can obtain information about the health status of lithium batteries.
This enables non-professional users to accurately assess the health status of lithium batteries through simple means, thus increasing their adoption rate.
Smart Images
Figure CN121114804A_ABST
Abstract
Description
Technical Field
[0001] This invention, a learning-based lithium battery health status classification method, belongs to the field of lithium battery performance evaluation technology. Background Technology
[0002] Lithium-ion batteries are used in many fields. With continued use, the internal resistance of lithium-ion batteries increases, and their capacity decreases, thus affecting their performance and safety. To describe the performance of lithium-ion batteries, the State of Health (SOH) metric has emerged. This metric is the ratio between the battery's current maximum usable capacity and its initial capacity, and it is crucial for understanding the degree of battery aging and remaining lifespan.
[0003] Currently, there are specialized methods to obtain battery health status, such as assessing state of health (SOH) by measuring the battery's actual capacity and internal resistance, or using characteristic parameters such as charging curves for modeling and analysis, combined with machine learning techniques for prediction. However, these methods require professional personnel to perform the testing, and their adoption is not widespread.
[0004] However, most lithium battery users are non-professionals, yet they are equally concerned about the health of the lithium batteries in their products. Therefore, there is a widespread demand for a simple method to roughly determine the health status of lithium batteries. Summary of the Invention
[0005] To address the aforementioned technical requirements, this invention proposes a learning-based lithium battery health status classification method. By learning, the key parameters of the lithium battery's health status are accurately obtained. Based on the uncertainty of whether it is healthy, a classification strategy is formulated, enabling the acquisition of the lithium battery's health status through non-professional parameters.
[0006] The objective of this invention is achieved as follows:
[0007] The learning-based lithium battery health status classification method includes the following steps:
[0008] Step a: Collect existing lithium battery data, including the service life categorized by short, medium and long life, the battery capacity categorized by large, medium and small size, whether it has been bumped or knocked, whether it is a brand name product, and whether it is in good condition.
[0009] Step b: Calculate the uncertainty of whether the lithium battery is healthy;
[0010] Step c: Calculate the uncertainty for each item under each condition;
[0011] Step d: Calculate the overall uncertainty for each item;
[0012] Step e: Develop a classification strategy based on the uncertainty of whether the lithium battery is healthy and the overall uncertainty of each item.
[0013] In the learning-based lithium battery health status classification method described above, the uncertainty calculation method for whether the lithium battery is healthy in step b is as follows:
[0014] -p 健康 log2p 健康 -p 不健康 log2p 不健康
[0015] In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
[0016] In the learning-based lithium battery health status classification method described above, in step c, the uncertainty of each item under each condition is calculated only for that specific item and condition, specifically regarding the uncertainty of whether the lithium battery is healthy. The calculation method is as follows:
[0017] -p 健康 log2p 健康 -p 不健康 log2p 不健康
[0018] In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
[0019] In the learning-based lithium battery health status classification method described above, the overall uncertainty of each item in step d is calculated by multiplying the uncertainty of each case by the probability of that case, and then summing the results.
[0020] In the learning-based lithium battery health status classification method described above, in step e, the item whose overall uncertainty differs the most from the uncertainty of whether the lithium battery is healthy obtained in step b is used as the first finite classification criterion.
[0021] The beneficial effect of the learning-based lithium battery health status classification method of this invention is that: by learning the key parameters of the lithium battery with accurate health status, and judging the uncertainty of whether it is healthy, a classification strategy is formulated, so as to obtain the health status of the lithium battery by acquiring non-professional parameters. Attached Figure Description
[0022] Figure 1 This is a flowchart of the learning-based lithium battery health status classification method of the present invention. Detailed Implementation
[0023] The specific embodiments of the present invention will now be described in further detail with reference to the accompanying drawings. Specific Implementation Method 1
[0025] The following is a theoretical implementation of the learning-based lithium battery health status classification method of this invention.
[0026] The flowchart of the learning-based lithium battery health status classification method in this specific implementation is as follows: Figure 1 As shown, it includes the following steps:
[0027] Step a: Collect existing lithium battery data, including the service life categorized by short, medium and long life, the battery capacity categorized by large, medium and small size, whether it has been bumped or knocked, whether it is a brand name product, and whether it is in good condition.
[0028] Step b: Calculate the uncertainty of whether the lithium battery is healthy;
[0029] Step c: Calculate the uncertainty for each item under each condition;
[0030] Step d: Calculate the overall uncertainty for each item;
[0031] Step e: Develop a classification strategy based on the uncertainty regarding the health of the lithium battery and the overall uncertainty of each item. In step b, the method for calculating the uncertainty regarding the health of the lithium battery is as follows:
[0032] -p 健康 log2p 健康 -p 不健康 log2p 不健康
[0033] In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
[0034] In step c, the uncertainty for each item under each condition is calculated only for that item and that condition, specifically regarding the uncertainty of whether the lithium battery is healthy. The calculation method is as follows:
[0035] -p 健康 log2p 健康 -p 不健康 log2p 不健康
[0036] In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
[0037] In step d, the overall uncertainty of each item is calculated by multiplying the uncertainty of each case by the probability of that case, and then summing the results.
[0038] In step e, the item whose overall uncertainty differs the most from the uncertainty about the health of the lithium battery obtained in step b will be used as the basis for finite classification. Specific Implementation Method Two
[0040] The following is a simulation data implementation of the learning-based lithium battery health status classification method of the present invention.
[0041] The lithium battery health status classification method based on learning in this specific implementation method yields the following lithium battery health data (step a):
[0042] Table 1 Lithium Battery Health Data Table
[0043]
[0044] The table shows a total of 14 sets of lithium battery health data, with 5 "no" and 9 "yes". Therefore, the uncertainty regarding the health of the lithium battery is (step b):
[0045]
[0046] Let's look at the first item, service life. There are three possibilities: short, medium, and long. Taking short as an example, it appeared 5 times. Three of these appeared as unhealthy conditions, and two appeared as healthy conditions. Therefore, the uncertainty of short service life is:
[0047]
[0048] Similarly, the uncertainty under the service life item is 0, and the uncertainty under the service life item is 0.971.
[0049] The overall uncertainty of the service life item is:
[0050]
[0051] The difference between the uncertainty of whether the lithium battery is healthy and the uncertainty of whether the battery is healthy is 0.246. Similarly, the difference of the overall uncertainty of the battery capacity item is 0.029, the difference of the overall uncertainty of the impact item is 0.151, and the difference of the overall uncertainty of the brand name item is 0.048. Step e is now complete.
[0052] Since the difference between the service life item and the uncertainty of lithium battery health is the largest, we first classify them according to service life, dividing Table 1 into three sub-tables, as follows:
[0053] Table 2 Health data of lithium batteries under short service life
[0054]
[0055] Table 3 Health data of lithium batteries after years of use
[0056]
[0057] Table 4. Health data of lithium batteries under long service life.
[0058]
[0059]
[0060] Each sub-table can be further subdivided using the previous method, ultimately yielding a unique health status based on usage years, battery capacity, whether it has been bumped or damaged, and whether it is a brand name.
Claims
1. A learning-based lithium battery health status classification method, characterized in that, Includes the following steps: Step a: Collect existing lithium battery data, including the service life categorized by short, medium and long life, the battery capacity categorized by large, medium and small size, whether it has been bumped or knocked, whether it is a brand name product, and whether it is in good condition. Step b: Calculate the uncertainty of whether the lithium battery is healthy; Step c: Calculate the uncertainty for each item under each condition; Step d: Calculate the overall uncertainty for each item; Step e: Develop a classification strategy based on the uncertainty of whether the lithium battery is healthy and the overall uncertainty of each item.
2. The learning-based lithium battery health status classification method according to claim 1, characterized in that, In step b, the method for calculating the uncertainty of whether the lithium battery is healthy is as follows: -p 健康 log2p 健康 -p 不健康 log2p 不健康 In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
3. The learning-based lithium battery health status classification method according to claim 1, characterized in that, In step c, the uncertainty for each item under each condition is calculated only for that item and that condition, specifically regarding the uncertainty of whether the lithium battery is healthy. The calculation method is as follows: -p 健康 log2p 健康 -p 不健康 log2p 不健康 In the formula, p 健康 p represents the probability of a lithium battery being healthy. 不健康 This indicates the probability that the lithium battery is unhealthy.
4. The learning-based lithium battery health status classification method according to claim 1, characterized in that, In step d, the overall uncertainty of each item is calculated by multiplying the uncertainty of each case by the probability of that case, and then summing the results.
5. The learning-based lithium battery health status classification method according to claim 1, characterized in that, In step e, the item whose overall uncertainty differs the most from the uncertainty about the health of the lithium battery obtained in step b will be used as the basis for finite classification.