Fuzzy Performance Indicators for the Control of Manufacturing Processes

Abstract

This study shows one use of the fuzzy subset theory with the aim of taking into account the qualitative, the imprecise and/or uncertain knowledge in a manufacturing context. The focus is more particularly on the evaluation aspects of the industrial processes and activities. This evaluation concerns the results of the different activities or processes’ enactments and consists in comparing them to the assigned objectives. One means to effect this comparison is given by performance indicators. Usual indicators only treat precise and numerical data, while the objectives can be imprecisely, subjectively or gradually defined and the measures uncertainly expressed. By using the fuzzy, the possibilistic and the probabilistic formalisms, the proposed indicators deal with both numerical and symbolic data, and provide either a performance measure or a performance evaluation. Moreover, always in contrast with the usual evaluation, the proposed indicators evaluate a relative performance, with regard to the objective on the one hand, and to the real conditions of the considered enactments on the other hand. While the absolute performance is related to the efficiency measure, the relative performance is useful for the control of the processes. The evaluation of these performances is also based on some fuzzy extensions of general concepts of correspondence, proximity, distance...These ideas are applied to some problems encountered in one ski production process. This production involves first an assembly activity whose performance is driven by many parameters; and at the end of the production, a control quality activity which is performed by human operators, on the basis of subjective features.