Battery Cell Anomaly Detection Using Reduced Voltage Extremes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current battery management systems face challenges in efficiently detecting and predicting anomalies in battery systems due to the large volumes of data generated, making resource-saving and long-term monitoring uneconomical, leading to suboptimal battery operation and potential failures like battery fires.
Innovation Solution
A method utilizing a reduced battery cell element measurement data set, comprising only the maximum and minimum voltage values of battery cells, which is transmitted and analyzed to detect anomalies, allowing for early prediction and prevention of battery system issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comprehensive individual analysis of each battery cell is performed, then anomaly detection precision is improved, but resource consumption and data processing complexity increase significantly
Solution Approach 1:
The patent segments the battery system into individual cell groups and analyzes them separately using unsupervised learning models. This allows precise monitoring of each segment while avoiding the need to process all cell data comprehensively, thus maintaining detection precision while reducing overall complexity
Solution Approach 2:
The patent introduces intermediate features extracted from battery data (such as voltage differences, temperature gradients, and current patterns) that serve as mediators between raw data and anomaly detection. These intermediate representations capture essential information while reducing data dimensionality and processing complexity
2Reliability
If all battery cell measurement data is stored and processed, then detection reliability is improved, but storage requirements and processing resources increase excessively
Solution Approach 1:
The patent extracts only the most relevant features from the complete battery measurement data, such as voltage differences between cells, temperature variations, and current patterns. This extraction process retains the essential information needed for reliable anomaly detection while dramatically reducing the volume of data that needs to be stored and processed
Solution Approach 2:
The patent transforms raw battery measurement data into derived parameters and features (e.g., voltage gradients, temperature rates of change, capacity fade indicators) that are more informative for anomaly detection. This parameter transformation maintains detection reliability while reducing data redundancy and storage requirements
3Reliability
If long-term monitoring of all battery cells is implemented, then early anomaly prediction is improved, but economic feasibility deteriorates due to high resource consumption
Solution Approach 1:
The patent applies partial action by implementing unsupervised learning models that focus on detecting anomalies in cell groups rather than performing comprehensive analysis on all cells continuously. This approach provides sufficient early anomaly prediction capability while significantly improving resource efficiency and economic feasibility
Solution Approach 2:
The patent employs unsupervised learning models that automatically learn normal battery behavior patterns and detect deviations without requiring continuous human intervention or extensive labeled training data. This self-service capability enables long-term monitoring with reduced operational resources and improved economic feasibility
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a computer program and a method for detecting and/or predicting anomalies in a battery system comprising two or more battery cell elements, comprising the following steps: providing a reduced battery cell element measurement data set, which was derived on the basis of a battery cell element measurement data set comprising, for each battery cell element of the battery system, one or more first state variables, in particular temperature, current flow and/or voltage and/or first state variables described by combinations thereof, of the respective battery cell element with respect to one or more time points as measurement data, such that the reduced battery cell element measurement data set contains, for a first state variable and for a time point: a first, selected according to a first criterion from the set of measurement data of the battery cell elements from the battery cell element measurement data set,for this time and this first state variable, selected measurement date, as well as a second, according to a second criterion from the set of measurement data of the battery cell elements from the battery cell element measurement dataset, for this time and this first state variable, selected measurement date, includes determining and/or predicting an anomaly of the battery system based on the reduced battery cell element measurement dataset, or a dataset based thereon.