Battery State-of-Health Trajectory Modeling for Unknown Cell Types
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the state of health of electrical energy stores, such as vehicle batteries, are inaccurate due to individual usage patterns and varying battery types, especially for unknown battery types where specific electrochemical models cannot be applied, leading to substantial model errors and limited insight into cell chemistry and internal structure.
Innovation Solution
A computer-implemented method that continually provides operating variables to an empirical state of health model, parameterizes a trajectory function using multiple state of health points based on time-dependent reference variables, and provides this function to predict the state of health trajectory, allowing for improved determination and prediction of the state of health, even for unknown battery types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional state of health models are used for unknown battery types, then the model implementation is simple, but the measurement precision deteriorates due to substantial model errors
Solution Approach 1:
The system performs preliminary actions by collecting and storing operating variables and state of health data during normal battery operation before the actual state of health assessment is needed. This preliminary data collection enables the subsequent machine learning model to be trained and parameterized without requiring complex real-time measurements or modifications to the battery itself, thus maintaining simplicity while improving accuracy
Solution Approach 2:
The patent introduces an intermediary machine learning model that acts as a mediator between the raw operating variables and the state of health determination. This intermediary layer processes the data through learned patterns and relationships, enabling accurate state of health assessment for unknown battery types without requiring direct application of conventional models that are type-specific
2Device complexity
If empirical state of health models are used without considering individual usage patterns, then the device complexity remains low, but the reliability deteriorates due to inaccuracy in predicting state of health
Solution Approach 1:
The system implements feedback by continuously monitoring operating variables and comparing actual state of health measurements with predictions from the machine learning model. The model is retrained and refined using this feedback loop, incorporating actual usage patterns and outcomes to improve future predictions. This feedback mechanism enhances reliability while keeping the overall system architecture relatively simple
Solution Approach 2:
The patent applies parameter changes by adapting the machine learning model parameters based on individual usage patterns observed during battery operation. Instead of using fixed parameters from conventional models, the system dynamically adjusts parameters to reflect actual usage conditions, thereby improving reliability without requiring a complete overhaul of the model structure
3Ease of operation
If conventional state of health models are applied to batteries with varying usage patterns, then the ease of operation is maintained, but the measurement precision worsens due to inability to account for usage-individual operating characteristics
Solution Approach 1:
The machine learning model achieves universality by being applicable to multiple battery types and usage patterns without requiring type-specific customization. The model learns general patterns from training data that transfer across different battery chemistries and operating conditions, enabling accurate state of health determination for unknown battery types while maintaining ease of operation across diverse applications
Data Source
AI summary
A computer-implemented method for ascertaining a state of health trajectory of an electrical energy store of a device, in particular an electrically driveable motor vehicle. The method including continually providing characteristics of operating variables that characterize operation of the electric energy store, ascertaining states of health based on the operating variables using an empirical state of health model at multiple times, parameterizing a trajectory function for describing the state of health trajectory based on multiple state of health points that each indicate one of the states of health of the ascertained states of health based on a time-dependent reference variable, and providing the trajectory function.