Battery Detection via Curve Fitting and Step-Curvature Radius Analysis

The patent pre-establishes a database of electrical properties for healthy batteries under various conditions (temperature, charge state, aging) before actual operation. This preliminary data preparation enables rapid comparison during runtime without requiring long-term monitoring, thus resolving the contradiction between detection reliability and detection time.

The patent segments the battery operation into distinct states (charging, discharging, resting) and temperature ranges, creating separate characteristic curves for each segment. By comparing real-time data against pre-established segment-specific curves, the system achieves rapid and accurate anomaly detection without needing continuous long-term monitoring.

The patent divides the battery characteristic curve into multiple segments (charging phase, discharging phase, resting phase) and applies specific fitting methods to each segment. This segmentation allows the use of simpler, targeted fitting approaches for each phase rather than attempting a single complex fit across the entire cycle, thereby maintaining high accuracy while reducing overall computational complexity.

Different fitting methods are applied to different segments of the characteristic curve based on their specific characteristics. For example, polynomial fitting may be used for smooth charging phases while other methods are applied to discharging phases with different behavior. This local optimization of fitting strategies improves detection accuracy without requiring uniformly complex computation across all data points.