Search

International audience; This paper is concerned by the industrial application of Ordered Weighted Averaging OWA operators, introduced by Yager. These operators allow to express the type of compromises, by the notion of linguistic quantifiers, such as "most" of criteria. The interest of this method of aggregation is, beyond its simplicity of use, its evaluation of products according a unique scale. Furthermore, the weights are not fixed by criteria but by levels of performance. In this paper we present a methodology of classification of products by two approaches. The first one is based on a learning sample and the second one on linguistic quantifiers. An industrial application, from a food production, illustrates these approaches. We then discuss the classifications obtained by these two approaches and we present a comparison.

This study deals with the development of an assisting tool, which helps the decision-making taking into account the human aspect of the decision-maker preferences. The proposed method is applied to the multicriteria optimisation of a reactive extrusion process.

To develop product within shorter deadlines remains a major challenge for the businesses. One of the possible answers is to capitalize "ready to use" expert knowledge. The choices of the customer being multiple, our research is oriented towards tolls and methods analysis and optimisation multi-criteria. In this paper, a generic approach and the Rough Sets Method (RSM) are briefly presented, which help the decision-making taking into account the human aspect of the decision maker preferences. This allows informal knowledge to be formalized based on the "preferences" and "no preferences" of experts. RSM is applied to an industrial study in a high yield pulping process and cheese production.

This study deals with the development of an assisting tool, which helps the decision-making taking into account the human aspect of the decision maker preferences. The proposed method is applied to the multicriteria optimisation of a reactive extrusion process.

Some of the most practical aspects of operations research to have appeared in the last decade are multiple decision criteria methodologies. With these new methods, conflicting objectives can be formally incorporated more naturally into the decision process and those solutions that offer the best compromise can be identified. These techniques can be used with advantage in engineering. This paper describes how these decision-making tools can help to optimise bioprocess operations. One particular optimisation algorithm, the net flow method, is described and used to determine the optimal operating condition for the production of gluconic acid. It is shown that these new tools, in addition to providing guidance to optimize the fermentation, also provide valuable information regarding the zone of possible solutions.

Multicriteria optimization of processes without particular knowledge is obviously of pratical interest. The usual way of dealing with such problems is based on choice of a structure for the model, which can be difficult without previous knowledge on the process. Consequently, this classical way does not seem to work properly in such conditions. This work describes an easy and wide using iterative methodology to solve the problem of the determination of an optimal zone of Pareto in which the decision maker picks up the working conditions corresponding to his choices. This new approach relies on the only available data, variation intervals for each variable and allows to minimize the number of experiments. This new method proves to be very accurate for the definition of an optimal working zone for the synthesis of a biosurfactant from lactose.

The use of an extended Kalman filter is useful and performant, however, it has has proven great handicap in preseence of important inaccuracy about the model. A new formulation of the PID filter is proposed that allow to overcome the Kalman filter dcfcticncies. Moreover, taking the constraints as artificial measurements is shown to be an enhancement of the “observability” of the overall system. We show also trough simulations the promising role of the Generalized Predictive Control in the optimisation of biomass productivity.