Hybrid Battery State Sensor with Dynamic Parameter Updating
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional battery state estimation methods, both model-based and data-driven, face challenges due to the complexity of battery processes, leading to reduced accuracy over time, especially in lithium-ion batteries, where parameters change frequently, making initial models and mappings less accurate.
Innovation Solution
A hybrid sensor approach is developed, combining model-based and data-driven methods to estimate battery state by updating model parameters using measured charge data, with methods like extremum-seeking for model-free learning, allowing for concurrent updates of multiple parameters and simplifying the model to reduce computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model-based methods are used to estimate battery state, then the estimation can be performed using physical models, but the accuracy reduces over time as battery parameters change
Solution Approach 1:
The patent implements dynamic parameter updating by continuously adapting model parameters based on real-time battery measurements. The system transitions from static initial models to dynamic models that evolve with battery aging and operating conditions, maintaining accuracy over extended periods through ongoing calibration.
Solution Approach 2:
The system employs feedback mechanisms where estimated battery states and measurements are continuously compared, and the differences are used to adjust model parameters. This closed-loop approach ensures the model adapts to actual battery behavior, compensating for parameter drift and maintaining estimation accuracy.
2Measurement precision
If data-driven methods are used to directly estimate battery state, then ground truth data can enable accurate initial mapping, but the training process becomes complex and requires extensive data
Solution Approach 1:
The patent segments the estimation problem into two parts: using data-driven methods only for parameter estimation while relying on simplified physical models for state estimation. This division reduces the complexity of the overall system compared to using complex data-driven models for direct state estimation, while still leveraging the accuracy benefits of data-driven approaches where they are most effective.
Solution Approach 2:
The system introduces model parameters as an intermediary between measurements and state estimation. Instead of directly mapping measurements to states using complex data-driven models, the approach uses measurements to update parameters, which then feed into simpler physical models for state estimation, reducing overall system complexity.
3Reliability
If complex models representing chemical processes are used, then the model can capture battery behavior, but the computational requirements increase
Solution Approach 1:
The patent changes the approach from using complex structural models to using simplified models with adaptive parameters. By focusing on parameter adaptation rather than model complexity, the system maintains accurate battery behavior representation while significantly reducing computational requirements and energy consumption.
Data Source
AI summary
A method for estimating a state of a battery determines, using a sensor, physical quantities of the battery indicative of a charge of the battery to produce a measured charge of the battery. The method also estimates the physical quantities of the battery using a model of the battery stored in a memory to produce an estimated charge of the battery and updates at least one parameter of the model of the battery to reduce a difference between the measured charge of the battery and the estimated charge of the battery. The method determines the state of the battery using the updated model of the battery.