Self-error learning framework-based algorithm for parameter recovery of extended Wiener-Hammerstein systems subject...

Area of Science:

• Control Systems Engineering
• Nonlinear System Identification
• Signal Processing

Background:

• Accurate system identification is crucial for control and analysis.
• Hysteresis nonlinearity and quantized measurements pose significant challenges.
• Existing estimation algorithms may lack performance in complex scenarios.

Purpose of the Study:

• To propose a novel estimation scheme for extended Wiener-Hammerstein systems.
• To address parameter identification under hysteresis nonlinearity and quantized measurements.
• To enhance estimation performance compared to traditional methods.

Main Methods:

• Development of a self-error learning framework.
• Introduction of an adaptive filter for data extraction from contaminated signals.
• Derivation of an identification error expression using filtered data.
• Proposal of an online compensation estimation error variable.
• Design of a new adaptive law with adaptive recursive gain.

Main Results:

• Effective extraction of useful identification data from noisy measurements.
• Elimination of regression vector effects on convergence performance.
• Online verification of persistent excitation (PE) condition for regressors.
• Strict proof of estimator convergence under general PE conditions.
• Demonstrated effectiveness through two examples and a real-world case.

Conclusions:

• The proposed estimation scheme offers high-performance parameter identification for complex systems.
• The self-error learning framework effectively handles hysteresis nonlinearity and quantized data.
• The method shows significant improvements over classic estimation algorithms.