Battery SOH Prediction Using Outlier-Excluded Averaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current battery management systems face challenges in accurately predicting the state-of-health (SOH) of batteries, leading to errors in charging and discharging strategies, which affects the efficiency and longevity of hybrid and electric vehicles.
Innovation Solution
A method and device that estimate SOH values, create a history table of these values, and determine a targeted SOH by averaging candidate values excluding the maximum and minimum values, using an SOH estimation unit, data storage unit, and processing unit, potentially with EEPROM storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If maximum and minimum SOH estimation values are included in the average calculation, then more data points are utilized, but prediction accuracy decreases due to outlier errors
Solution Approach 1:
The patent extracts and removes maximum and minimum outlier values from the SOH estimation data set before calculating the average. This is achieved by storing multiple SOH estimation values in a history table, identifying the maximum and minimum values, and excluding them from the average calculation to prevent their distortion effect on the final SOH prediction accuracy.
2Measurement precision
If only average SOH values are used, then prediction accuracy improves, but the system cannot capture extreme battery conditions
Solution Approach 1:
The patent applies different processing treatments to different portions of the SOH estimation data. The maximum and minimum values receive special treatment (exclusion from average calculation), while the intermediate values are used for the average SOH prediction. This local differentiation allows the system to maintain high prediction accuracy for normal conditions while still monitoring extreme values separately for safety and adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed are a method and a device for predicting a state-of-health of a battery and a battery management system using same. The method for predicting a state-of-health of a battery includes estimating state-of-health (SOH) estimation values for predicting an aged degree of the battery for each time; aligning the plurality of estimated values to create a history table; extracting candidate values from the history table; and determining a targeted SOH of the battery based on the candidate values.