Battery Degradation Forecasting With Hidden Markov Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for monitoring and predicting the degradation of rechargeable batteries in electric vehicles are limited by the accuracy of measurement instruments, which are often designed for short-term monitoring and are prone to noise and interference, leading to uncertain and noisy measurement data.
Innovation Solution
A method utilizing a trained hidden Markov model (HMM) to improve the accuracy of battery degradation state estimation by accounting for uncertainties in measurement data and physical models, allowing for probabilistic predictions of degradation progression and remaining usable life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If measurement instruments designed for short-term monitoring are used, then device complexity is reduced, but measurement precision deteriorates due to noise and interference
Solution Approach 1:
The patent introduces a trained Hidden Markov Model as an intermediary between the noisy measurement data and the degradation state determination. The HMM processes the noisy observations from simple measurement instruments and probabilistically infers the true degradation state, effectively mediating between low-quality measurements and accurate degradation assessment without requiring complex measurement hardware
Solution Approach 2:
The patent replaces complex high-precision measurement instruments with a computational approach using Hidden Markov Models. Instead of improving the physical measurement system, the solution substitutes a mathematical/probabilistic system that processes the noisy data to extract accurate degradation information, achieving high measurement precision through computation rather than hardware complexity
2Measurement precision
If higher-quality measuring instruments are used, then measurement precision improves, but cost increases
Solution Approach 1:
The patent employs inexpensive measurement instruments that can be easily manufactured and deployed, accepting that individual measurement data points may be noisy or inaccurate. The system uses multiple such inexpensive measurements over time, processed through the HMM, to achieve accurate degradation assessment without investing in expensive high-precision instruments
Solution Approach 2:
The trained Hidden Markov Model serves as a cost-effective intermediary that enhances the capability of cheap measurement instruments. By processing multiple noisy measurements through the HMM's probabilistic framework, the system achieves degradation measurement accuracy comparable to expensive instruments while maintaining low hardware costs
3Reliability
If multiple measurement parameters are monitored, then reliability of degradation determination improves, but device complexity increases
Solution Approach 1:
The Hidden Markov Model serves as a universal processing framework that can handle multiple different measurement parameters (voltage, current, temperature, charge-discharge cycles) through a unified probabilistic model. This multi-functional approach allows the system to reliably determine degradation state using various measurement types without requiring separate complex processing systems for each parameter
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Method (100) for determining an approximation and/or prediction (2a) for the true degradation state (2) of a rechargeable battery (1) comprising the steps: • a time series (3) of measured values (2b) of the degradation state determined at points in time in the past, discretized in predefined time steps, is provided (110); • a trained Hidden Markov Model, HMM (4), is provided (120) which, depending on the true degradation state (2), indicates: o the probability with which value (2b) of the degradation state is observed during the measurement, and o the probability with which this true degradation state (2) is maintained for how long, and/or the probability with which this true degradation state transitions into a worse degradation state (2') in the next time step;• From the observed time series (3) and the HMM (4), the most probable course (2*) of the true degradation state (2) in the past is determined (130), which is consistent with the observed time series (3); • From the most probable course (2*), the desired approximation and/or forecast (2a) is evaluated (140).;