ML Model Selection for Battery SOC Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery management systems face challenges in accurately estimating state of charge (SOC) and state of health (SOH) values, leading to inefficient battery management and potential safety issues such as battery-related fires.
Innovation Solution
The implementation of a machine learning (ML) model in battery management systems to enhance the estimation of SOC and SOH values, allowing for adaptive and dynamic battery management decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional SOC estimation methods (chemical method, voltage method, current integration method) are used, then the battery management system can provide basic SOC estimation, but the accuracy is compromised over battery age and the methods are non-adaptive to dynamic situations
Solution Approach 1:
The system dynamically selects different ML models based on battery conditions (SOH value, temperature, charge/discharge rate). The model selection module chooses from multiple pre-trained ML models (e.g., LSTM, GRU, CNN) depending on the current operating state, making the SOC estimation adaptive to dynamic situations while maintaining high accuracy across different battery ages and conditions.
2Adaptability or versatility
If multiple ML models are maintained for different battery conditions, then the adaptability to dynamic situations is improved, but the device complexity increases
Solution Approach 1:
The system automatically selects the appropriate ML model based on input features (SOH value, temperature, charge/discharge rate) without requiring manual intervention. The model selection module compares current battery conditions against predefined thresholds and automatically chooses the most suitable pre-trained model, reducing operational complexity while maintaining support for multiple battery types and conditions.
Solution Approach 2:
Multiple ML models are pre-trained offline for different battery conditions and states of health. This preliminary preparation allows the system to have ready-to-use models for various scenarios, eliminating the need for real-time model training or complex runtime model generation, thus reducing device complexity while maintaining high adaptability.
3Ease of operation
If SOC is estimated from other measurements and known parameters, then the battery management system can operate without direct SOC measurement, but errors occur in SOC estimation leading to inefficient or improper battery management
Solution Approach 1:
Traditional mathematical estimation methods (voltage method, current integration method) are replaced with machine learning-based estimation. The ML models learn complex non-linear relationships between measurable parameters (voltage, current, temperature, SOH) and SOC from training data, providing more accurate SOC estimation without requiring direct SOC measurement. This substitution significantly improves measurement precision while maintaining ease of operation.
Data Source
AI summary
One example method for using a machine learning (ML) model in battery management comprises: receiving one or more selection inputs for selecting an ML model for providing battery management information, wherein the one or more selection inputs include a state of health (SOH) value associated with a battery system; selecting, using the selection inputs, the ML model from a plurality of ML models; obtaining, using model inputs and the ML model, the battery management information associated with the battery system; and performing, using the battery management information, a battery management decision for managing the battery system.