Battery Health Estimation Using Real-Time Voltage Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the state of health of electric or hybrid vehicle batteries are not reliable or adaptable for real-time, online conditions due to complex physico-chemical aging phenomena and require prior estimates of battery state charge and resistance, leading to biased results and divergence over time.
Innovation Solution
A method and system that estimate battery state of health directly from real-time signals using sensors for voltage, current, temperature, and vehicle speed, constructing models through pattern extraction and comparison, employing dynamic time warping and statistical methods to account for variations in usage conditions, allowing for continuous and accurate health assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If equivalent circuit modeling is used to estimate battery state of health, then the estimation can be performed in real-time, but the model parameters must be adapted for each battery technology and require prior estimates of state of charge and resistance which introduce biases
Solution Approach 1:
The patent extracts and removes the problematic intermediate variables (state of charge and resistance estimates) from the estimation chain. By directly comparing voltage patterns at different aging levels without requiring these intermediate estimates, the method eliminates the source of biases while maintaining real-time capability.
Solution Approach 2:
The patent creates a universal estimation method that works across different battery technologies without requiring technology-specific model adaptation. The pattern comparison approach is general and can be applied to various battery types, eliminating the need for separate equivalent circuit models for each technology.
2Measurement precision
If equivalent circuit parameters are adapted for each battery technology, then measurement precision may improve, but the methodology loses flexibility and requires complex prior knowledge
Solution Approach 1:
The patent achieves universal applicability by using pattern comparison that works across different battery technologies. The method compares voltage patterns directly without requiring technology-specific parameter adaptation, thus maintaining both precision and adaptability.
Solution Approach 2:
The patent changes the estimation approach from parameter-based (requiring adaptation of circuit parameters) to pattern-based (comparing voltage evolution patterns). This parameter transformation enables the method to work universally across different battery technologies without losing precision.
3Ease of manufacture
If prior estimates of state of charge and resistance are used as input variables, then the model can be constructed, but biases are introduced that cause divergence of results over time
Solution Approach 1:
The patent removes the biased intermediate estimates (state of charge and resistance) from the input variables. By directly comparing voltage patterns without these intermediate steps, the method eliminates the introduction of biases while still enabling model construction through pattern matching.
4Manufacturing precision
If tests in controlled conditions are used to build the model, then manufacturing precision may be improved, but the model becomes unrepresentative of real usage conditions
Solution Approach 1:
The patent performs preliminary pattern extraction from controlled test conditions, then uses these patterns as reference for comparison during real usage. This preliminary action in controlled conditions enables accurate modeling while the subsequent real-time comparison adapts to actual usage conditions.
Solution Approach 2:
The method uses feedback by continuously comparing real-time voltage patterns against reference patterns obtained from controlled tests. This feedback mechanism allows the model to remain accurate in real conditions by adapting to actual usage through pattern matching.
Data Source
Figure 1~3B
Figure 2
Figure 4~5B
AI summary
Disclosed is a method for estimating the state of health of a battery in an electric or hybrid vehicle during operation thereof, said method comprising the following steps: a) during operation of the battery, acquiring a time series of measurements of the speed (v) or acceleration of the vehicle and simultaneously at least one time series of measurements (I, U, P) of a variable selected from among a current or a power supplied by the battery and a voltage at the battery terminals; b) extracting segments of said time series which correspond to speed or acceleration patterns that meet at least one predefined condition; and c) establishing estimations of the state of health of the battery by applying at least one continuous estimation model or classification model to said segments of the time series. Also disclosed are a device and a system for carrying out said method as well as a method for creating a continuous estimation model or classification model of said type.