Battery State Estimation Using Feature-Space Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current battery state estimation methods are inefficient in accurately determining the state of a battery, particularly in real-time, due to the complexity of processing high-dimensional sensing data and the need for extensive calibration and pattern recognition, which can lead to delayed or inaccurate fault detection.
Innovation Solution
A battery state estimation apparatus and method that acquires sensing data, segments it based on time intervals, corrects time errors, maps the data to a predetermined feature space using a reference matrix, and compares the feature vector to predefined patterns to determine the battery state, including normal, abnormal, and fault states, with the ability to update reference information for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-dimensional sensing data is processed using traditional pattern recognition methods, then comprehensive battery state analysis is achieved, but processing time increases and real-time estimation is delayed
Solution Approach 1:
The sensing data processing is divided into distinct segments: raw data acquisition, feature extraction, mapping to feature space, and pattern comparison. This segmentation allows each step to be optimized independently, with feature extraction reducing data dimensionality before the computationally intensive pattern recognition step, thereby reducing overall processing time while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary mapping of sensing data to a predetermined feature space using a reference matrix before pattern recognition. This preprocessing step transforms high-dimensional data into a lower-dimensional feature space that preserves essential characteristics, reducing the computational burden of subsequent pattern matching and enabling faster real-time estimation.
2Measurement precision
If extensive calibration and pattern recognition are performed, then battery state estimation accuracy is improved, but operational complexity increases
Solution Approach 1:
The patent extracts only the essential features from high-dimensional sensing data by mapping to a predetermined feature space using a reference matrix. This extraction process isolates the most relevant characteristics needed for pattern recognition, reducing operational complexity by eliminating redundant data processing while maintaining estimation accuracy.
Solution Approach 2:
The patent transforms the sensing data from its original high-dimensional parameter space to a lower-dimensional feature space through linear mapping. This parameter transformation reduces the number of variables that need to be processed during pattern recognition, simplifying the operational complexity while preserving the essential information needed for accurate battery state estimation.
3Speed
If high-dimensional sensing data is processed in real-time, then timely fault detection is achieved, but computational resources are overwhelmed
Solution Approach 1:
The patent projects sensing data from high-dimensional space to a lower-dimensional predetermined feature space using a reference matrix. This dimensionality reduction transforms the computational problem from handling high-dimensional data to working with a smaller feature vector, significantly reducing computational complexity while enabling real-time processing speed for timely fault detection.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A battery state estimation apparatus includes a sensing data acquirer configured to acquire sensing data on a battery, and a battery state estimator configured to approximate the sensing data by mapping the sensing data to a predetermined feature space, and compare the approximated sensing data to predetermined reference information to estimate a state of the battery.