Propulsion Battery RUL Prediction Using Aging Mode Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the remaining useful lifetime of vehicle propulsion batteries lack accuracy due to insufficient consideration of different aging mechanisms, leading to inadequate planning and maintenance of battery systems.
Innovation Solution
A method involving constructing a voltage feature vector from charging events, mapping it to a battery state of health model with aging mode distribution, and using a physical aging model to predict the remaining useful lifetime, taking into account various aging mechanisms like active material loss and solid electrolyte interface growth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional state of health monitoring methods are used, then the monitoring process is simple, but the prediction accuracy of remaining useful lifetime is insufficient
Solution Approach 1:
The patent segments the battery aging process into multiple distinct aging modes (e.g., lithium plating, SEI growth, electrolyte decomposition) and models each mode separately. This segmentation allows the system to capture different degradation mechanisms with appropriate physics-based models for each, thereby improving prediction accuracy while keeping each individual model manageable in complexity
Solution Approach 2:
The patent transforms the prediction approach by changing from monitoring single aggregate parameters (like overall state of health) to tracking multiple mode-specific parameters (such as lithium plating rate, SEI growth rate). This parameter transformation enables more precise characterization of different aging mechanisms and their contributions to overall battery degradation
2Reliability
If multiple aging mechanisms are considered, then the remaining useful lifetime prediction becomes more reliable, but the computational complexity increases
Solution Approach 1:
The patent implements a dynamic framework where the relative importance of different aging modes evolves over time based on operating conditions. The system continuously updates the contribution of each aging mechanism (lithium plating, SEI growth, etc.) as the battery undergoes different charge/discharge cycles, temperatures, and states of charge, allowing adaptive prediction that reflects real-time battery behavior
Solution Approach 2:
The patent introduces an intermediary layer that connects electrical measurements (voltage, current, temperature) to physical aging mechanisms. This intermediary translation layer uses physics-based relationships to convert easily measurable electrical parameters into estimates of underlying chemical degradation processes, bridging the gap between simple measurements and complex aging mechanisms
3Loss of information
If voltage features from charging events are used, then the state of health parameters can be determined, but additional data processing is required
Solution Approach 1:
The patent performs preliminary processing of voltage data during charging events by extracting relevant features (voltage gradients, plateaus, reversibility metrics) as the charging occurs. This real-time feature extraction during the charging process prepares the data for subsequent aging mode analysis without requiring separate dedicated measurement campaigns or post-processing of entire charge cycles
Solution Approach 2:
The patent leverages the battery's own charging process to generate the diagnostic information needed. Normal charging voltage measurements, which are already being taken for battery management, are repurposed to extract aging indicators. The system transforms routine charging data into valuable degradation information without requiring additional external testing or specialized measurement equipment
Data Source
AI summary
A method for determining a remaining useful lifetime of a propulsion battery of a vehicle. The method includes constructing a voltage feature vector from voltage data of a charging event for the battery, mapping the voltage feature vector to a battery state of health model including an aging mode distribution of different battery aging mechanisms to determine state of health parameters and the present aging mode distribution of the battery; predicting the remaining useful lifetime (RUL) using a physical aging model with the state of health parameters and the present aging mode distribution as inputs to the physical aging model.