Battery SOC Observer Covariance Matrix Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the state-of-charge (SOC) of batteries, particularly in aircraft batteries, rely on empirical and intuitive definitions of the process noise covariance matrix Q, which affects the reliability of SOC estimation due to lack of consideration for actual operating conditions.
Innovation Solution
A method to determine the process noise covariance matrix Q by calculating voltage differences and impedance values at various operating points, using the method of least squares, and storing these values based on state-of-charge, current, and temperature to produce a covariance matrix that accounts for specific operating conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the process noise covariance matrix Q is defined empirically or intuitively, then the initialization is simplified, but the reliability of SOC estimation deteriorates due to lack of consideration for actual operating conditions
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing the process noise covariance matrix Q for multiple operating points before actual battery operation. During runtime, the system only needs to select the pre-computed Q matrix corresponding to the current operating point, avoiding real-time complex calculations while ensuring reliability through operating-condition-specific values.
Solution Approach 2:
The patent changes the parameter Q from a fixed empirical value to a variable that depends on operating conditions (temperature, charge/discharge rate, SOC range). By computing Q as a function of these parameters offline and storing in lookup tables, the system achieves both ease of implementation and improved reliability through condition-adaptive values.
2Reliability
If the process noise covariance matrix Q is defined to account for operating conditions, then the reliability of SOC estimation is improved, but the device complexity increases due to multiple operating points and storage requirements
Solution Approach 1:
The patent segments the continuous operating space into discrete operating points based on temperature ranges, charge/discharge rates, and SOC intervals. Each segment has its own pre-computed Q matrix stored in lookup tables. This segmentation transforms a complex continuous optimization problem into manageable discrete segments, reducing real-time computational complexity while maintaining reliability.
Solution Approach 2:
The complex task of computing Q matrices for various operating conditions is performed in advance during system setup or calibration. The results are stored in memory for quick retrieval during operation. This preliminary computation eliminates the need for complex real-time calculations, reducing device complexity during actual battery monitoring while preserving reliability through accurate, condition-specific Q values.
3Ease of manufacture
If fixed initialization values are used for matrices P, R and Q, then the implementation is simplified, but the accuracy of SOC estimation deteriorates because actual operating conditions are not taken into account
Solution Approach 1:
The patent transforms the fixed initialization approach into a parameter-adaptive approach. Instead of using constant Q values, the system selects Q matrices based on current operating parameters (temperature, current rate, SOC range). This parameter change enables the system to maintain implementation simplicity through lookup table selection while achieving high accuracy through condition-matched Q values.
Solution Approach 2:
The patent applies local quality by using different Q matrix values for different operating conditions rather than a single global value. Each operating point (defined by temperature range, charge/discharge rate, and SOC interval) has its own optimized Q matrix. This local optimization improves accuracy for each specific operating condition while maintaining simplicity through pre-computed lookup tables.
Data Source
AI summary
This method for determining a process noise covariance matrix for tuning an observer of the state-of-charge of a battery of electrical accumulators includes the steps of:determining, for each operating point of a set of operating points of the battery, a value of at least one component (R0, Z1) of an electrical model of the battery;obtaining, for each of said values, a determination error (r(SOCk)) for said component;storing the values of said component and the determination errors;calculating the standard deviation (σ[r(SOC1 . . . p)]) for the determination errors for various operating points of the battery; andproducing the covariance matrix (Q) on the basis of the calculated standard deviation.