A synergistic architecture for adaptive, intelligent control system development

Over the previous decade, a new field of investigation termed computational intelligence (CI) has emerged and begun to flourish. This field is primarily concerned with the development of computer systems that are capable of adapting to and exploiting information about their environments, much like organisms in natural systems are capable of doing. It is no coincidence therefore, that the field of CI relies heavily on computer techniques patterned after natural systems. Many of these techniques including neural networks, genetic algorithms, and fuzzy logic have demonstrated their utility in solving problems independent of other methods. However, as the systems we seek to control, design, and improve become increasingly complex, it is likely that no single CI technique will be adequate. This paper describes an architecture combining the three CI techniques listed above that can be used to produce process control systems suitable for effectively manipulating complex engineering systems characterized by relatively slow process dynamics. Implementation of the architecture results in a level-two intelligent control system. The effectiveness of the level-two intelligent controller is demonstrated via application to an operating phosphate processing plant