Experimental and Theoretical Techniques for Nonlinear Identification and Adaptive Cancellation
Abstract
Under this project we developed and demonstrated new techniques for identification and adaptive control. For linear identification, we developed quadratically constrained least squares (QCLS) identification, which extends classical least squares identification and is more resistant to the effects of system noise. For nonlinear identification, we developed a sequential method for combined Hammerstein-nonlinear feedback models. This method uses a piecewise linear approximation to the system nonlinearity and is based on computationally efficient numerical procedures. For adaptive control, we developed controllers for adaptive disturbance rejection and adaptive stabilization. These methods were implemented on laboratory experiments involving active noise control and rotating imbalance stabilization.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 05, 2000
- Accession Number
- ADA385627
Entities
People
- Dennis S. Bernstein
Organizations
- University of Michigan