Battery Kalman Filter Covariance Monitoring for Stable Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional battery management systems using Kalman filters for estimating battery cell states and parameters often become unstable due to non-deterministic behavior, leading to divergent and inaccurate parameter estimation, particularly in dynamic battery cell parameter changes like internal resistance.
Innovation Solution
A method and device employing a dual Kalman filter with noise component determination and adaptation, comparing covariance behavior to desired levels, and automatically deactivating the filter when instability is detected to maintain stability, allowing for dynamic and adaptive estimation of battery cell states and parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a Kalman filter is used to estimate battery cell states and parameters, then the estimation capability is improved, but the system becomes unstable and divergent over time
Solution Approach 1:
The patent implements a stability monitoring mechanism that continuously observes the Kalman filter's covariance behavior and provides feedback to detect instability. When divergence is detected through covariance analysis, the system triggers corrective actions including switching to a backup filter or deactivating the adaptive parameter estimation, thereby maintaining system reliability while preserving estimation accuracy during stable operation.
Solution Approach 2:
The patent introduces dynamic switching capability between different filter configurations and operational modes. The system can transition from adaptive parameter estimation mode to fixed parameter mode or switch between primary and backup Kalman filters based on real-time stability assessment, allowing the system to adapt its behavior to maintain both accuracy and stability under varying conditions.
2Adaptability or versatility
If adaptive parameter estimation is implemented in the Kalman filter, then the battery cell parameter observation capability is improved, but the system becomes non-deterministic and unstable
Solution Approach 1:
The stability monitoring mechanism continuously evaluates the covariance behavior of adaptive parameter estimates and provides feedback to detect when adaptability causes instability. When parameter estimation leads to divergent behavior, the system automatically deactivates the adaptive mode and switches to deterministic fixed parameter operation, thereby maintaining reliability while preserving adaptability during stable conditions.
Solution Approach 2:
The patent implements dynamic parameter switching between adaptive and fixed configurations. The system monitors covariance parameters to determine when adaptive parameter changes are causing instability, and automatically reverts to fixed parameter values, allowing the system to utilize adaptability benefits when safe while maintaining determinism when stability is compromised.
3Device complexity
If the Kalman filter operates without stability monitoring, then the device complexity is reduced, but the system becomes unstable after indefinite time
Solution Approach 1:
The patent incorporates a lightweight stability monitoring mechanism that continuously observes filter covariance behavior with minimal computational overhead. This feedback system detects instability early through covariance analysis and triggers corrective actions, providing long-term stability assurance without significantly increasing device complexity.
Solution Approach 2:
The system performs preliminary stability assessment through covariance monitoring before instability fully develops. By detecting divergent behavior early in its development through continuous covariance evaluation, the system can take preventive corrective actions rather than waiting for complete failure, thereby ensuring long-term reliability without requiring complex post-failure recovery mechanisms.
Data Source
AI summary
Method for monitoring a stable convergence behaviour of a Kalman filter (KF), which estimates states and/or parameters of an energy storage system, in particular a battery cell (BZ), wherein a covariance behaviour—provided by the Kalman filter (KF)—in terms of at least one state and/or parameter of the energy storage system is compared with a corresponding desired covariance behaviour of the state and/or parameter, wherein the Kalman filter (KF) is automatically deactivated for each state and/or each parameter of the energy storage system, of which the covariance behaviour exceeds the corresponding desired covariance behaviour.