An intelligent systems approach for manufacturing process analysis and control system design

The main objective of this thesis is to develop Intelligent Systems Methodologies for modeling nonlinear systems, recognizing inverse patterns, and designing intelligent control systems. Specifically, this study presents a non-conventional approach to modeling a manufacturing process subsystem, followed by further research on inverse models by Artificial Neural Networks; and at the final stage, different control system schemes are proposed to apply system knowledge to implement inverse control in a fuzzy model of the manufacturing process.;The major contributions of the research in this thesis include the following: (1) The use of fuzzy logic theory to model a dynamic manufacturing process system. (2) Proof that Artificial Neural Networks (ANNs) can be employed to approximate complex nonlinear industrial process systems. (3) The discovery that Artificial Neural Networks to be used on the applications are strictly decided by the information that is provided for the ANN training. If the information that is provided is adequate to produce a unique solution in the ANN training input -- output direction, an ANN will form as a function approximator. When the information is reduced in the ANN training session, the uniqueness will disappear in the ANN training input -- output direction; in this case, an ANN will form as well, but can only work for mapping this input -- output pattern. (4) The discovery that virtual input functions and virtual output functions are a good practical approach in solving inverse Neural Network Control with non -- uniqueness. Furthermore, this will also enable the connecting of the computer simulation to the real manufacturing process modeling and will also enhance the controlling ability. (5) The complete design of two novel inverse controller programs for this specific manufacturing process, which can perform a human operator's job and achieve the best controlling goals.