A simplified fuzzy multivariable structure in a manufacturing environment

Applications of fuzzy control to industrial processes are mainly of multivariable structure. Using the traditional Zadeh principle would require a multidimensional relation to be developed representing a fuzzy model of the system. Such a multidimensional relation would result in memory overload to most industrial computers. Moreover, it would result in a highly complex compositional rule of inference to achieve the final output(s) of the system. This paper proposes a new simplified technique that avoids such complexity as well as memory overload for multivariable structure. Section 2 describes the proposed simplified multivariable technique to avoid memory overload. Section 3 demonstrates these techniques in the form of a robotic welding example where the objective is to control the speed of a robotic arm following an irregular path of weld. The speed value is dependent on the cavity size and determined by the cavity width and cavity depth as inputs. Section 4 describes an experimental application of the technique applied to an industrial process in the manufacture of force transducers termed as the ‘cornering process’. This application is composed of a two-inputs-two-outputs system.