Battery Life Prediction From Spliced Decay Curve Segments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing battery life prediction methods, such as electrochemical principle prediction and genetic algorithm prediction, are computationally intensive and prone to large prediction errors, and fail to accurately explain the decay pattern of battery state of health.
Innovation Solution
A battery data processing method that involves obtaining preset life decay curves for different life decay types, extracting curve segments based on predicted operating conditions, and splicing these segments end-to-end to create a predicted life decay curve, reducing prediction errors by considering the decay pattern under various conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If electrochemical principle prediction is used to predict battery life, then battery performance can be optimized by improving battery material design, but the calculation is complex and computationally intensive
Solution Approach 1:
The patent segments the continuous battery life decay curve into multiple discrete curve segments corresponding to different life decay types (e.g., calendar aging, cycle aging, fast charging aging). Each segment is independently characterized by key parameters such as start/end points, decay rates, and inflection points. This segmentation simplifies the complex electrochemical degradation process into manageable discrete units that can be processed and predicted more efficiently.
Solution Approach 2:
The patent transforms the complex continuous decay curve into a set of discrete parameters including decay rates, curve segment boundaries, and characteristic points. By changing the representation from continuous function to discrete parameters, the computational complexity is reduced while preserving the essential degradation characteristics needed for accurate life prediction.
2Productivity
If genetic algorithm prediction is used to predict battery life, then the prediction is more efficient compared to electrochemical principle prediction, but the prediction error is large and it fails to explain the decay pattern
Solution Approach 1:
The patent introduces curve segment parameters as intermediary elements between the raw battery test data and the final life prediction. These parameters (decay rates, inflection points, segment boundaries) serve as mediators that capture the essential degradation patterns while enabling efficient computation. This intermediary layer preserves the explanatory power of electrochemical models while achieving the computational efficiency of data-driven approaches.
Solution Approach 2:
The patent performs preliminary analysis to identify and extract key curve segment parameters from battery degradation data before conducting the actual life prediction. By pre-processing the data to extract meaningful parameters such as decay rates and characteristic points, the method prepares the information in advance, enabling faster and more accurate predictions without requiring complex real-time computations.
3Reliability
If traditional prediction methods are used, then battery life can be predicted, but the prediction error is large due to failure to consider decay patterns under different life decay types
Solution Approach 1:
The patent applies different decay characteristics and parameters to different segments of the battery life curve, recognizing that the degradation mechanism varies at different stages. Each curve segment is characterized by local quality parameters specific to that decay type (e.g., different decay rates for calendar vs. cycle aging). This localized approach improves prediction accuracy by capturing the non-uniform nature of battery degradation throughout its lifecycle.
Data Source
Figure 1
Figure 2A~2C
Figure 3
AI summary
Provided are a battery data processing method, an electronic device, and a storage medium. The method includes: obtaining (101) preset life decay curves of a target battery under multiple different life decay types; extracting (103) one curve segment from a respective one of the preset life decay curves; splicing (104) multiple curve segments end-to-end to obtain a predicted life decay curve of the target battery; and determining (105) a predicted life of the target battery based on the predicted life decay curve.