An expert system prototype assisting the statistical validation of simulation models

The validation stage is often neglected in a simulation project, due to the high level of knowledge which is necessary. In this paper, the expert system approach is proposed to assist the simulation user in selecting an appropriate validation method, application of this method in the appropriate way, and interpreting the results. The main objective of such a system is to regroup validation techniques, scattered in literature, into the same knowledge base, and then to complete it with knowledge in statistics, simulation, and validation techniques, coming from experts. This study is part of a more general study on an Intelligent Environment for Simulation. Due to the difficulty in designing such a validation expert system, the application domain is restricted to the validation of observable discrete systems (particularly manufacturing systems), and to methods based on statistical tests. The functions of the expert system and the chosen knowledge representation are outlined and a prototype developed using K.E.S. software, is introduced. A brief example is given which illustrates the system capabilities. Finally the conclusion states that the expert system approach is adapted, due to restrictions which were imposed.