Battery Pack SOC Estimation with Adaptive Thevenin-AEKF Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional SOC estimation methods for lithium-ion batteries in new energy vehicles suffer from inaccuracies due to fixed initial parameters, poor adaptability to varying conditions, and reliance on offline data, leading to reduced accuracy and stability in real-time SOC estimation.
Innovation Solution
A method utilizing a Thevenin model and a noise adaptive extended Kalman filter (AEKF) algorithm with a time-varying forgetting factor (VFF-RLS) to dynamically adjust parameters, incorporating real-time data for improved SOC estimation accuracy, by fitting an OCV-SOC function and jointly determining maximum and minimum SOCs based on adaptive noise covariance adjustment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed initial parameters are used in conventional SOC estimation methods, then the algorithm implementation is simple, but the adaptability to varying battery conditions deteriorates
Solution Approach 1:
The patent applies dynamics by transforming fixed initial parameters into time-varying adaptive parameters. The forgetting factor λ(k) dynamically adjusts the weight of historical data, allowing the algorithm to adapt to changing battery conditions while maintaining computational feasibility through recursive updates.
Solution Approach 2:
The patent changes parameters by introducing a time-varying forgetting factor λ(k) that evolves with battery operating conditions. This parameter adaptation enables the algorithm to respond to battery aging, temperature variations, and load changes, resolving the contradiction between simplicity and adaptability.
2Loss of time
If offline test data is used to identify battery model parameters, then the test period and sample requirements are reduced, but the SOC estimation accuracy under real-time varying conditions deteriorates
Solution Approach 1:
The patent applies preliminary action by performing offline parameter identification to obtain initial battery model parameters, then using these as starting points for online adaptive refinement. This two-stage approach reduces the overall test period while maintaining accuracy through real-time adaptation.
Solution Approach 2:
The patent implements feedback by continuously comparing estimated SOC with actual battery behavior and using the forgetting factor to adjust parameter weights. This feedback mechanism maintains high SOC estimation accuracy under varying real-time conditions while keeping offline testing requirements manageable.
3Use of energy by moving object
If conventional extended Kalman filter algorithm with fixed parameters is used, then the computational load is low, but the tracking performance under dynamic operation conditions deteriorates
Solution Approach 1:
The patent applies dynamics by replacing fixed parameters with time-varying adaptive parameters in the extended Kalman filter. The forgetting factor λ(k) enables the algorithm to track changing battery conditions dynamically while maintaining computational efficiency through recursive least squares updates.
Solution Approach 2:
The patent changes parameters by introducing adaptive parameter adjustment through the forgetting factor mechanism. This allows the filter to maintain low computational load while significantly improving tracking performance under dynamic operation conditions through parameter evolution.
4Measurement precision
If data-based methods such as neural networks are used for SOC estimation, then the SOC estimation accuracy may be improved, but the requirement for vast sample data for training makes implementation difficult
Solution Approach 1:
The patent applies this principle by using simple, computationally efficient recursive algorithms with adaptive parameters instead of complex neural networks. The forgetting factor mechanism provides a lightweight alternative that achieves good accuracy without requiring vast training datasets or complex computational infrastructure.
Solution Approach 2:
The patent changes parameters by using adaptive parameter adjustment through the forgetting factor rather than fixed parameters from extensive training. This approach achieves competitive SOC estimation accuracy with significantly reduced implementation complexity and data requirements.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Disclosed is a method and apparatus for estimating the SOC of a battery, a device, a battery module, and a storage medium. The method includes: fitting a relationship between open-circuit voltage (OCV) and the state of charge (SOC) based on the OCV corresponding to an SOC range of 0 to 100% to obtain an OCV-SOC function; creating a Thevenin model of a tested battery pack based on equivalent circuit parameters and the OCV-SOC function; inputting real-time parameters acquired from a to-be-tested battery pack into the Thevenin model for detection according to a preset VFF-RLS algorithm with a time-varying forgetting factor; jointly determining a maximum SOC and a minimum SOC based on a preset noise adaptive extended Kalman filter (AEKF) algorithm and the Thevenin model; and determining an overall SOC of the to-be-tested battery pack according to the maximum SOC and the minimum SOC. In this way, the accuracy for estimating the overall SOC can be improved.