A novel recursive learning estimation algorithm of Wiener systems with quantized observations.

Area of Science:

• Control Systems Engineering
• Signal Processing
• Machine Learning

Background:

• Accurate parameter estimation is crucial for control system performance.
• Wiener systems with quantized outputs present unique identification challenges.
• Existing methods may struggle with noise and limited precision in quantized data.

Purpose of the Study:

• To develop a novel recursive learning identification algorithm for Wiener systems with quantized output.
• To improve the precision and convergence rate of parameter estimation.
• To validate the proposed algorithm through simulations and real-world experiments.

Main Methods:

• A recursive learning identification approach is proposed.
• Adaptive filtering is used for data preprocessing.
• Novel filtered and intermediate variables are developed to derive estimation error.
• A new loss function is established to enhance precision and convergence.
• Minimization of the loss function yields a recursive learning estimator with improved gain performance.

Main Results:

• The proposed algorithm effectively estimates parameters of Wiener systems with quantized output.
• The estimation error converges to zero under continuous excitation conditions.
• Illustrative examples and a real-life experiment demonstrate the algorithm's efficiency and accuracy.

Conclusions:

• The novel recursive learning approach provides an effective solution for identifying Wiener systems with quantized outputs.
• The method enhances parameter estimation accuracy and convergence speed.
• The algorithm's practical applicability is confirmed through experimental validation.