Battery SOH Prediction Using ML Ensemble Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating the state-of-health (SOH) of batteries are expensive, time-consuming, and require lab measurements, which are not practical for widespread use.
Innovation Solution
A prediction module uses a multi-part process involving machine learning models to predict future SOH values of batteries by training an estimation model with lab-measured data and then using historical SOH data to train an ensemble of models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional electrochemical lab measurements are used to evaluate battery SOH, then measurement precision is improved, but loss of time and cost increase significantly
Solution Approach 1:
The patent creates a digital copy of the battery's SOH state through machine learning models that replicate the behavior of expensive lab measurements. The estimation model generates historical SOH data that mirrors what would be obtained from actual electrochemical tests, allowing virtual replication of measurement results without physical lab testing.
Solution Approach 2:
The patent replaces the mechanical/chemical lab measurement system with a computational machine learning system. Instead of using electrochemical impedance spectroscopy and other physical measurement techniques in laboratories, the system uses trained ML models that process battery data to estimate SOH, substituting physical measurement mechanisms with information processing mechanisms.
2Measurement precision
If traditional electrochemical lab measurements are used to evaluate battery SOH, then measurement precision is improved, but cost increases significantly
Solution Approach 1:
The patent uses inexpensive computational resources and readily available battery data to replace expensive lab measurement processes. The machine learning models run on standard computing infrastructure, processing data that already exists from normal battery operation, thereby eliminating the need for costly specialized lab equipment and services.
Solution Approach 2:
The system creates accurate virtual copies of lab measurement results through the estimation model, which generates historical SOH data that reflects what actual measurements would show. This digital copying approach eliminates the need for repeated expensive physical measurements while maintaining measurement precision.
3Productivity
If digital modeling and simulating of battery states are used, then productivity is improved, but measurement precision deteriorates due to reliance on physical condition knowledge
Solution Approach 1:
The patent implements feedback mechanisms where the machine learning models are trained on actual lab-measured SOH values and continuously refined based on comparison between predicted and actual results. The system uses loss functions to measure prediction errors and updates model parameters to minimize these errors, creating a self-improving system that maintains high precision while achieving fast evaluation.
Solution Approach 2:
The patent performs preliminary training of machine learning models using extensive battery data and lab measurements before deployment. This preliminary action creates pre-trained models that can quickly evaluate SOH without needing detailed physical condition knowledge during actual operation, combining the speed of digital modeling with the precision of lab measurements.
Data Source
AI summary
A system for determining a future state-of-health (SOH) of a battery includes a memory communicably coupled to a processor and storing instructions that, when executed by the processor, cause the processor to acquire battery data of a battery for which a future SOH is to be predicted. The instructions also cause the processor to identify a selected model from an ensemble of models according to at least a time to the future SOH and a time length of the battery data. The instructions further cause the processor to predict the future SOH using the battery data as an input to the selected model.