Battery SOH Monitoring Using Normalized Rate-of-Change Inputs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the State of Health (SoH) of battery systems are either impractical, require extensive processing power, or provide relative rather than absolute measurements, limiting their suitability for real-world implementation, especially in embedded systems.
Innovation Solution
A monitoring system that includes sensors to measure battery characteristics during charge or discharge operations and a processing device with a preprocessor to determine normalized rates of change of these characteristics. A shallow neural network then uses this preprocessed data to accurately determine the SoH of the battery system, reducing computational burden and making it feasible for embedded systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a deep neural network is used for SoH prediction, then prediction accuracy is improved, but processing power requirements increase beyond embedded system capabilities
Solution Approach 1:
The patent segments the neural network into a shallow architecture with fewer layers and nodes, dividing the computational task into simpler stages. This segmentation reduces the overall processing power requirements while maintaining acceptable SoH prediction accuracy, making it feasible for embedded systems with limited computational resources.
Solution Approach 2:
The patent changes key parameters of the neural network including reducing the number of layers, nodes per layer, and training iterations. These parameter modifications directly reduce computational complexity and processing power requirements while preserving the network's ability to provide accurate SoH predictions for battery systems.
2Measurement precision
If a full slow charge/discharge cycle is performed to measure charge capacity, then SoH measurement accuracy is improved, but operational time and battery lifetime are reduced
Solution Approach 1:
The patent applies partial action by using a shallow neural network that requires fewer computational steps and processing cycles compared to a full deep neural network. This partial computational action achieves acceptable SoH prediction accuracy without the excessive processing time and resource consumption, enabling practical implementation in embedded systems.
3Measurement precision
If model-based methods are used to determine charge capacity, then measurement capability is improved, but decoupling from other state variables becomes difficult
Solution Approach 1:
The patent segments the complex battery system modeling into a simplified shallow neural network structure with distinct input, hidden, and output layers. This segmentation allows the network to process multiple state variables (voltage, current, temperature) separately before integrating them for SoH prediction, reducing the difficulty of decoupling interactions between variables while maintaining measurement accuracy.
Data Source
AI summary
A monitoring system (120) for a battery system (110) comprising one or more battery cells (110). Monitoring system (120) comprises one or more sensors (125) for measuring characteristics associated with battery system (110) during a charge or discharge operation, and a processing device (130) communicatively coupled with the one or more sensors (125). Processing device (130) comprises a preprocessor (135) configured to receive measurement data from the at least one sensor (125) and a neural network (140) configured to receive processed data from preprocessor (135). Based on the received measurement data, preprocessor (135) determines a normalized rate of change of a first measured characteristic against one of: time; or a second measured characteristic associated with battery system (110) measured during the charge or discharge operation, wherein the second measured characteristic is different from the first measured characteristic. Neural network (140) is configured to use the normalized determined rate of change from preprocessor (135) as an input to determine a state of health (SOH) of battery system (110).