Battery Defect Cause Classification Using Electrical Test Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting battery defects during manufacturing processes can only identify the defect state but not determine the cause, requiring manual disassembly and analysis, which is time-consuming and prone to variations in accuracy among workers.
Innovation Solution
A defect type classifying system that includes a measurer to record electrical characteristics, a converter to generate input data from measured voltage data, and a machine-learning based defect type predicting module to classify the defect type based on learning data, reducing the need for manual disassembly and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual disassembly and analysis is used to determine defect causes, then defect causes can be identified, but the process is time-consuming and prone to accuracy variations
Solution Approach 1:
The patent replaces the manual mechanical disassembly and analysis process with an automated measurement system that uses gauges to collect electrical characteristics (voltage, current, resistance) and a machine learning model to determine defect causes. This substitution eliminates the time-consuming manual disassembly while providing consistent, objective analysis results based on measured data patterns.
Solution Approach 2:
The system performs preliminary measurements of electrical characteristics during the manufacturing process itself, before defects fully manifest. By collecting and analyzing voltage, current, and resistance data in advance, the system can identify potential defect causes early, reducing the need for time-consuming post-manufacturing disassembly and analysis.
2Reliability
If manual disassembly is performed to find defect causes, then defect information can be obtained, but worker skill variations cause accuracy deviations
Solution Approach 1:
The patent replaces the unreliable manual analysis process with an automated system that uses standardized gauge measurements and a trained machine learning model. This substitution ensures that defect analysis results are consistent and reliable, independent of individual worker skills or experience levels, as the system objectively processes measured electrical characteristics through the trained model.
Solution Approach 2:
The system performs self-analysis by automatically processing measured electrical characteristics through the machine learning model to determine defect causes. The measured data itself contains the information needed for diagnosis, and the system extracts this information autonomously without requiring external expert intervention, ensuring consistent results across different instances.
3Loss of information
If only defect state is detected by existing gauge methods, then measurement is simple, but defect cause information is lost
Solution Approach 1:
The patent makes the measurement system multi-functional by using the same gauge infrastructure to both detect defect states and determine defect causes. The system collects comprehensive electrical characteristics (voltage, current, resistance) that serve dual purposes: identifying that a defect exists and analyzing the measured patterns to determine the specific cause, eliminating the need for separate diagnostic procedures.
Solution Approach 2:
The machine learning model acts as an intermediary that transforms raw measured data into meaningful defect cause information. The model receives electrical characteristic measurements and intermediate processing results, then outputs definitive defect cause classifications. This intermediary layer extracts hidden information from the measured data that would be difficult to obtain through direct manual analysis alone.
Data Source
AI summary
A defect type classifying system includes: a measurer connected to a battery from which a defect type is detected, and measuring at least one of electrical characteristics of the battery for a predetermined test period and generating measured data; a converter for generating input data by converting the measured data; and a defect type predicting module for machine-learning learning data, and determining a defect type of the battery based on the input data. The input data may be appropriate for an input node of the defect type predicting module, and the defect type may be classified according to a cause of a short-circuit defect of the battery.