Li-Ion Battery Pack SoH Estimation for Heterogeneous Cells
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Estimating State-of-Health (SoH) parameters for battery packs with hundreds or thousands of interconnected cells is challenging due to scalability, computational costs, limited sensing, and complex dynamics, particularly because existing methods either require extensive data sets or are difficult to identify and interpret, and often lump cells together, losing heterogeneity information.
Innovation Solution
A method and system using sparse Gaussian process regression to estimate electrical parameter bounds as a function of State of Charge (SoC) and temperature, intelligently polling representative cells and employing recursive least squares for single cell parameter estimation, providing scalable and accurate SoH estimation across the battery pack.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physics-based electrochemical models are used for SoH estimation, then descriptive accuracy is improved, but model identification difficulty increases due to large number of parameters
Solution Approach 1:
The patent extracts only the essential electrical parameters (resistances and capacitances) from the full electrochemical model that are most critical for SoH estimation. This selective extraction maintains descriptive accuracy while reducing the number of parameters that need to be identified, thereby simplifying model identification complexity.
Solution Approach 2:
The patent segments the battery pack into individual cell models, each with its own electrical parameters. This segmentation allows for independent parameter estimation for each cell while capturing cell heterogeneity, making the overall identification process more manageable compared to treating the entire pack as a single complex system.
2Productivity
If equivalent circuit models are used for SoH estimation, then computational simplicity is improved, but parameter identification accuracy deteriorates due to nonlinear curve fitting requirements
Solution Approach 1:
The patent performs preliminary identification of electrical parameters at specific operating points (selected SoC and temperature conditions) before using these parameters for real-time SoH estimation. This preliminary action separates the complex identification task from the real-time estimation task, improving computational efficiency while maintaining accuracy through the use of pre-identified parameters.
Solution Approach 2:
The patent accounts for parameter changes by identifying electrical parameters at multiple operating points (different SoC and temperature levels) and selecting appropriate parameters based on current operating conditions. This approach maintains parameter identification accuracy across varying conditions while keeping the computational model simple for real-time use.
3Measurement precision
If all cells in battery pack are monitored individually, then SoH estimation accuracy is improved, but computational burden increases due to hundreds or thousands of cells
Solution Approach 1:
The patent applies partial action by estimating parameters for all cells but using a reduced set of representative cells for actual SoH estimation. This approach maintains accuracy by capturing cell heterogeneity through the representative subset while significantly reducing the computational burden compared to processing all cells individually in real-time.
Solution Approach 2:
The patent merges cells with similar characteristics into representative cell models. By grouping cells and using representative models for each group, the system maintains accurate SoH estimation that accounts for cell heterogeneity while reducing the computational load from processing thousands of individual cells to processing a manageable number of representative models.
4Measurement precision
If battery operation is stopped for experimental SoH procedures, then measurement accuracy is improved, but operational time is reduced
Solution Approach 1:
The patent enables continuous SoH estimation by using the equivalent circuit model approach that can operate during normal battery charging and discharging. The electrical parameters are identified from regular operational data without requiring the battery to be stopped or subjected to special experimental procedures, thus maintaining both accuracy and continuous operation.
Solution Approach 2:
The system uses the battery's own operational data (voltage, current, temperature during normal charging/discharging) to estimate SoH parameters. This self-service approach eliminates the need for external experimental procedures or battery stoppage, allowing accurate SoH measurement to be obtained from the battery's normal operational behavior itself.
Data Source
AI summary
A method for assessing a state of health of a battery having a plurality of heterogeneous cells includes subjecting the cells of the battery to a plurality of diagnostic current pulse cycles; identifying extreme cells based upon the cycles; estimating model parameters of the extreme cells; and estimating upper and lower bounds for the estimated model parameters. Estimating model parameters includes performing a recursive least squares analysis on the extreme cells. Estimating the upper and lower bounds for the estimated model parameters includes performing a sparse Gaussian process regression using the estimated model parameters.