Two-stage maximum likelihood weighted recursive least squares algorithm for nonlinear systems and an application in...

Area of Science:

• Control Systems Engineering
• Nonlinear System Identification
• Signal Processing

Background:

• Hammerstein systems are widely used in modeling complex nonlinear dynamics.
• Identifying parameters in systems with dead-zone nonlinearity and colored noise presents significant challenges.
• Existing methods may suffer from computational complexity or suboptimal performance.

Purpose of the Study:

• To develop robust parameter identification algorithms for Hammerstein systems with dead-zone nonlinearity.
• To address the challenge of colored noise interference in system identification.
• To improve estimation accuracy and computational efficiency compared to existing methods.

Main Methods:

• Introduction of a switching function to obtain a linear parameter expression for the piecewise nonlinear subsystem.
• Construction of a maximum likelihood weighted recursive least squares (ML-WRLS) identification algorithm based on the maximum likelihood principle.
• Proposal of a two-stage recursive least squares algorithm to enhance estimation accuracy and reduce computational load.

Main Results:

• The proposed two-stage ML-WRLS algorithm demonstrates improved steady-state performance over the standard ML-WRLS algorithm.
• The developed algorithms effectively identify parameters in Hammerstein systems with dead-zone nonlinearity and colored noise.
• Simulation examples validate the effectiveness and superiority of the proposed identification methods.

Conclusions:

• The novel two-stage recursive least squares algorithm offers a significant improvement for identifying Hammerstein systems with dead-zone nonlinearity.
• The proposed methods provide accurate and computationally efficient solutions for complex nonlinear system identification problems.
• This work contributes to the advancement of robust parameter estimation techniques in the presence of noise and nonlinearities.