Battery State Estimation Using Mixed SPKF and RLS Techniques
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current battery management systems for hybrid vehicles rely on inaccurate linear models for state and parameter estimation, and while non-linear models offer better performance, they are difficult to tune, leading to a need for improved estimation techniques.
Innovation Solution
A battery management system utilizing a mixed sigma-point Kalman filtering (SPKF) and recursive least squares (RLS) technique to estimate battery states and parameters, with the SPKF estimating states and RLS estimating parameters, and vice versa, to improve accuracy and tunability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If linear models are used for state and parameter estimation, then the system is easy to implement, but the estimation accuracy deteriorates
Solution Approach 1:
The patent segments the estimation process into two distinct parts: state estimation using SPKF and parameter estimation using RLS. This segmentation allows each algorithm to be optimized for its specific function while maintaining overall system simplicity. The state estimation handles dynamic battery states, while parameter estimation handles slow-varying battery parameters, resolving the contradiction between ease of implementation and estimation accuracy.
Solution Approach 2:
The patent creates a composite estimation system by combining SPKF and RLS algorithms into a unified framework. This composite approach leverages the strengths of both algorithms - SPKF's ability to handle non-linear state estimation and RLS's effectiveness in parameter identification - to achieve high accuracy while maintaining implementability through modular architecture.
2Measurement precision
If non-linear models are used for state and parameter estimation, then the estimation accuracy improves, but the system complexity and difficulty of tuning increase
Solution Approach 1:
The patent divides the complex non-linear estimation problem into manageable segments: SPKF handles the non-linear state estimation with its sigma-point approach, while RLS handles parameter estimation separately. This segmentation reduces overall system complexity by allowing each algorithm to focus on specific aspects of the estimation problem with dedicated tuning parameters.
Solution Approach 2:
The patent changes the approach to handling non-linearity by using SPKF's sigma-point transformation method instead of traditional linearization techniques. This parameter transformation approach maintains accuracy in non-linear regimes while keeping the computational structure manageable. The separate parameter estimation via RLS further simplifies tuning by focusing on a smaller set of battery parameters.
3Measurement precision
If non-linear models are used for state and parameter estimation, then the estimation accuracy improves, but the difficulty of tuning increases
Solution Approach 1:
The patent segments tuning into two distinct processes: SPKF tuning for state estimation dynamics and RLS tuning for parameter estimation. This segmentation simplifies the overall tuning difficulty by allowing engineers to adjust state and parameter estimation characteristics independently, rather than tuning a single complex non-linear model all at once.
Solution Approach 2:
The patent implements feedback mechanisms where parameter estimation results from RLS are fed into the SPKF state estimation, and state estimation results provide measurements for RLS parameter updates. This feedback loop enables automatic adaptation and reduces manual tuning effort, as the system self-adjusts based on observed battery behavior.
Data Source
AI summary
A battery management system for an electrified powertrain of a hybrid vehicle includes one or more sensors configured to measure voltage, current, and temperature for a battery system of the hybrid vehicle and a controller. The controller is configured to obtain an equivalent circuit model for the battery system, determine a set of states for the battery system to be estimated, determine a set of parameters for the battery system to be estimated, receive, from the sensor(s), the measured voltage, current, and temperature for the battery system, using the equivalent circuit model and the measured voltage, current, and temperature of the battery system, estimate the sets of states and parameters for the battery system using a mixed sigma-point Kalman filtering (SPKF) and recursive least squares (RLS) technique, and using the sets of estimated states/parameters for the battery system, control an electric motor of the electrified vehicle.