State-of-Health Modeling With Evolving Inter-Device Variability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting the state of health of batteries and fuel cells fail to accurately account for inter-device variability and the evolution of this variability over time, leading to inaccurate modeling of their degradation, which is crucial for determining optimal operation duration and maintenance schedules.
Innovation Solution
A method involving a computer-implemented model that combines Gaussian processes to represent both average evolution and inter-device variability, with a non-linear combination of random processes to capture the evolving variance, allowing for accurate characterization of the state of health over the device's lifetime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Gaussian process regression is used to predict battery state of health, then prediction capability is improved, but the model fails to accurately account for inter-device variability and its evolution over time
Solution Approach 1:
The model segments the state of health prediction into two distinct components: a deterministic mean evolution term and a stochastic variability term. This segmentation allows each component to be modeled separately with appropriate mathematical tools, enabling accurate capture of both average degradation trends and inter-device variability patterns.
Solution Approach 2:
The variability term is designed to be dynamic, with its variance evolving as a function of the mean state of health. This dynamic structure allows the model to adapt the variability characteristics according to the current degradation level, accurately representing how uncertainty changes throughout the device lifecycle.
2Reliability
If extensive aging tests are performed on batteries to quantify degradation, then degradation understanding is improved, but cost and time consumption increase significantly
Solution Approach 1:
The model performs preliminary characterization of degradation patterns and variability structures during the training phase using limited data. Once trained, the model can predict future degradation behavior without requiring continuous extensive testing, thus reducing ongoing time and resource requirements while maintaining reliable degradation understanding.
Solution Approach 2:
Instead of physically testing numerous batteries to understand degradation, the trained model creates a virtual copy of the degradation process that can be simulated and analyzed computationally. This digital twin approach allows extensive degradation analysis without the need for corresponding physical experiments.
3Device complexity
If the variance of inter-device variability is assumed constant, then model simplicity is improved, but accuracy in representing real degradation variability deteriorates
Solution Approach 1:
The model changes the parameter structure by making the variance of the variability term a function of the mean state of health rather than a constant. This parameter transformation allows the model to capture the evolving nature of degradation uncertainty while maintaining a relatively simple functional form that can be efficiently trained and applied.
Data Source
AI summary
A computer-implemented method for characterizing an evolution of the state of health of a population of devices with duration of operation comprises training a model on a database comprising, for each device among the population of devices, a value of duration of operation of the device and a corresponding state of health of the device, said model being a random process comprising at least a sum of: a term representing an average evolution of the state of health of the population of devices with duration of operation, and a term representing an inter-device variability of a degradation of the state of health at equal duration of operation, the variance of said term evolving with the duration of operation.