Battery State Estimation via Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery state estimation models require significant computational resources due to handling high-dimensional time-series data, making real-time estimation challenging for battery management systems with limited resources, and often include unnecessary information from complex chemical actions within batteries.
Innovation Solution
A battery state estimating apparatus that learns a battery state estimation model using discriminant analysis to transform high-dimensional data into a low-dimensional feature space, allowing for efficient pattern matching and estimation of battery states like SoC, SoH, and fault states by segmenting sensor signals and extracting feature patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high-dimensional time-series data model is used for battery state estimation, then the accuracy of state estimation is improved, but the computational complexity increases making real-time estimation difficult
Solution Approach 1:
The patent extracts only the essential features from high-dimensional time-series battery sensor data by identifying and removing unnecessary information related to complex chemical actions. This feature extraction process retains the critical patterns needed for accurate state estimation while significantly reducing the dimensionality and computational complexity of the data model.
Solution Approach 2:
The patent introduces an intermediary processing layer that transforms raw high-dimensional time-series sensor data into a simplified feature space. This intermediary representation maintains the essential information for accurate battery state estimation while being computationally efficient enough for real-time processing by battery management systems with limited resources.
2Loss of information
If a complex estimation model is used to capture all chemical actions in the battery, then the completeness of information is improved, but the computational resources required increase
Solution Approach 1:
The patent selectively extracts only the necessary information from complex battery sensor signals by identifying and removing redundant data related to unnecessary chemical actions. This extraction process maintains completeness of essential information for state estimation while eliminating computational waste on irrelevant chemical processes.
3Speed
If high-dimensional time-series data is processed in real-time, then the responsiveness of battery state estimation is improved, but the computational burden on the battery management system increases
Solution Approach 1:
The patent extracts essential features from time-series battery data in real-time by removing unnecessary information about complex chemical actions. This real-time feature extraction enables fast processing with reduced computational power requirements, allowing battery management systems to maintain responsiveness without excessive computational burden.
Data Source
AI summary
A battery state estimating apparatus includes a learner configured to learn a battery state estimation model comprising a label corresponding to a battery state; and a state estimator configured to estimate a battery state of a target battery using the battery state estimation model.