A fuzzy approach for the performance evaluation of manufacturing processes

This paper focuses on one of the problems related to industrial process performance, more particularly related to the evaluation aspect. The latter is provided by performance indicators which assess the results reached, according to pre-set objectives. The usual indicators only treat precise numerical data, while in a manufacturing context, the objectives can be flexible or graduated symbolic or subjective. Furthermore, the execution of the processes is often measured by physical sensors or human operators. Thus, the acquired information is eventually expressed with uncertainty. To deal with vagueness, imprecision, uncertainty, we propose to use some fuzzy approaches for the comparison of the measures with the assigned objectives, such as the fuzzy extension of the general concepts of correspondence and distance. On another hand and in contrast to the conventional evaluation, we also propose a relative performance. It results from the evaluation of the measures always with regard to the objective, but in accordance with the effective conditions of the execution of the process. Our propositions are illustrated by an application to some problems encountered in the control of a ski production process