THERE ARE many physical and socio-economic systems which are too complex, highly nonlinear, uncertain, incomplete or non-stationary, and have subtle and interactive exchanges with the operating environment, to make them nonviable for conventional controller synthesis. Yet such processes are successfully controlled by human skilled operators using substantially experiential knowledge; the operator may not know how or why he is controlling the system, but at least he knows what to do and is able to achieve complex performance criteria. Rather than mathematically model, the process under control, the operator, models in a heuristic manner. This alternate "top down" perspective in controller design requires the acquisition of heuristic and qualitative, rather than quantitative knowledge of expertise from the operator, process, and the means to represent it. Expert systems, neural nets and fuzzy logic are all non-model experiential methodologies for solving the above problem. Fuzzy logic controllers have been in development for over 20 years and have found widespread industrial application including autopilots, cement kilns, elevator controls, video camcorders, washing machine controls, air conditioners, TV contrast and brightness controls, car transmissions, antilock brakes, and local transportation. To facilitate this industrial application of fuzzy logic control there has been the development RISC type microprocessor architectures for carrying >50 kilo fuzzy logical operations per second.
Despite the advanced mathematical development of fuzzy logic theory, for single-input-output, multivariable processes, very few text books devoted to fuzzy logic have emerged. To remedy this shortfall, Witold Pedrycz, a notable