Document Type

Dissertation

Date of Award

Spring 5-31-1994

Degree Name

Doctor of Philosophy in Mechanical Engineering - (Ph.D.)

Department

Mechanical and Industrial Engineering

First Advisor

Rajesh N. Dave

Second Advisor

Nirwan Ansari

Third Advisor

Rong-Yaw Chen

Fourth Advisor

Zhiming Ji

Fifth Advisor

Nouri Levy

Abstract

The problem of adaptive nonlinear control, i.e. the control of nonlinear dynamic systems with unknown parameters, is considered. Current techniques usually assume that either the control system is linearizable or the type of nonlinearity is known. This results in poor control quality for many practical problems. Moreover, the control system design becomes too complex for a practicing engineer. The objective of this thesis is to provide a practical, systematic approach for solving the problem of identification and control of nonlinear systems with unknown parameters, when the explicit linear parametrization is either unknown or impossible.

Fuzzy logic (FL) and neural networks (NNs) have proven to be the tools for universal approximation, and hence are considered. However, FL requires expert knowledge and there is a lack of systematic procedures to design NNs for control. A hybrid technique, called fuzzy logic adaptive network (FLAN), which combines the structure of an FL controller with the learning aspects of the NNs is developed. FLAN is designed such that it is capable of both structure learning and parameter learning. Gradient descent based technique is utilized for the parameter learning in FLAN, and it is tested through a variety of simulated experiments in identification and control of nonlinear systems. The results indicate the success of FLAN in terms of accuracy of estimation, speed of convergence, insensitivity against a range of initial learning rates, robustness against sudden changes in the input as well as noise in the training data. The performance of FLAN is also compared with the techniques based on FL and NNs, as well as several hybrid techniques.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.