OBJECTIVE SATISFACTION ASSESSMENT USING NEURAL NETS FOR BALANCING MULTIPLE OBJECTIVES

In today's production systems, improving the use of the manufacturing resources and reacting efficiently to disturbances leads to schedules more and more adapted to the considered workshop. A generic software c an hardly take into account the specificity of each workshop: in that context, it is not sufficient anymore to provide a feasible schedule, and human expertise becomes necessary in order to improve the provided solution. This improvement requires the definition of synthetic perfo rmance indicators allowing us to assess a schedule before choosing imp rovement actions. Many performance indicators have been defined, howev er, they are seldom structured in order to supply a complete and progr essive assessment framework. We suggest in this paper a parametrable h ierarchic structure of performance indicators allowing us to aggregate the degree of satisfaction of elementary objectives, thus allowing th e definition of a compromise between these elementary objectives. Neur al networks have been tested in order to emulate the expertise involve d in the definition of such compromises. Neural networks enable us to express the satisfaction provided by a schedule in a synthetic way, th en to describe the satisfaction of the elementary objectives in order to select improvement actions. Using the same indicator values, severa l aggregation strategies can be considered and stored in order to adap t the assessment phase to the global situation of the workshop (e.g. i n the presence of overloads, under loads, rush orders, lateness, bottl enecks, etc.). The implementation of this method in an industrial sche duler, called IO, is in progress.