Identification of nonlinear Wiener-Hammerstein systems by a novel adaptive algorithm based on cost function...

Area of Science:

• Control Systems Engineering
• Nonlinear System Identification
• Signal Processing

Background:

• Nonlinear Wiener-Hammerstein systems are prevalent in various engineering applications.
• Accurate parameter identification is crucial for effective system control and analysis.
• System identification is often challenged by noise corruption in measured data.

Purpose of the Study:

• To develop an advanced parameter identification method for nonlinear Wiener-Hammerstein systems.
• To address challenges posed by noise in system data.
• To improve estimation accuracy and convergence speed compared to existing methods.

Main Methods:

• Exploiting filter gain to extract system data from noisy measurements.
• Developing an extended estimation error vector using auxiliary filtered variables.
• Proposing a novel cost function incorporating a discount term and an initial estimate penalty.
• Deriving an optimal parameter adaptive law based on the new cost function.

Main Results:

• The proposed algorithm demonstrates a faster convergence rate than conventional methods.
• Higher estimation accuracy is achieved using the novel cost function.
• Convergence analysis confirms that parameter estimation errors converge to zero.
• Effectiveness validated through simulations and experimental data from a turntable servo system.

Conclusions:

• The developed parameter identification scheme offers superior performance for nonlinear Wiener-Hammerstein systems.
• The novel cost function and filter gain approach effectively handle noisy data.
• The method is practical and validated for real-world applications.