Battery SOH Estimation With Attention-Based Trust Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurately estimating the state of health (SOH) of lithium-ion batteries is challenging due to various aging mechanisms influenced by operating and environmental conditions, which affect the internal state and capacity of batteries, necessitating continuous monitoring for safe operation.
Innovation Solution
A method and apparatus utilizing a pre-trained artificial neural network to estimate SOH by generating input data from battery parameters, calculating an attention score based on an attention map, and determining the reliability of the estimation value, incorporating an inverted bottleneck network structure for efficient memory use and accurate estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a pre-trained artificial neural network is used to estimate battery SOH, then estimation accuracy is improved, but computational complexity and memory requirements increase
Solution Approach 1:
The neural network model is pre-trained offline using extensive battery data and electrochemical models before deployment. This preliminary training phase allows the model to learn complex aging patterns and relationships, which are then stored as fixed weights in the deployed system, eliminating the need for complex real-time computations during actual SOH estimation
Solution Approach 2:
Instead of implementing complex electrochemical models directly in the battery management system, the patent creates a simplified copy (neural network) that replicates the behavior of the complex models. This neural network copy maintains high estimation accuracy while requiring significantly fewer computational resources for real-time operation
2Reliability
If real-time measurement of battery parameters is performed, then SOH estimation reliability is improved, but measurement and processing time increase
Solution Approach 1:
The patent extracts and utilizes only the most critical battery parameters (voltage, current, temperature) that have the highest impact on SOH estimation. By focusing on these key parameters rather than processing all available battery data, the system achieves reliable estimation with minimal processing time
Solution Approach 2:
The patent replaces complex mechanical and computational processing systems with an intelligent neural network system that can process measurements rapidly. The neural network's parallel processing capability allows it to evaluate multiple parameter relationships simultaneously, significantly reducing computation time while maintaining or improving reliability
3Measurement precision
If complex aging mechanisms are considered in SOH estimation, then estimation accuracy is improved, but system complexity increases
Solution Approach 1:
The patent merges multiple complex aging mechanisms (electrochemical reactions, material degradation, thermal effects) into a single integrated neural network model. This unified model learns the interactions between different aging mechanisms during training, allowing it to capture complex degradation patterns without requiring separate processing systems for each mechanism
Solution Approach 2:
The patent transforms complex physical and chemical aging parameters into neural network input features that the model can process efficiently. By changing the representation of aging mechanisms from detailed physical models to optimized neural network parameters, the system maintains accuracy while reducing complexity
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of estimating a state of health of a battery is performed by at least one computing device, and includes preparing a pre-trained artificial neural network, generating input data by measuring at least one parameter of the battery, inputting the input data into the pre-trained artificial neural network to obtain a health state estimation value of the battery and an attention map, calculating an attention score based on the attention map, and determining whether to trust the health state estimation value based on the attention score.