Optimization of a knowledge-based system by a meta-heuristic approach for the automotive diagnosis

Nowadays, one of the current challenges in the automotive industry is to establish a precise and accurate diagnosis in the after-sales workshops within a reasonable time in order to minimize the inconvenience for customers of the maintenance operations. The automotive industry had developed more and more electronic systems to enhance safety and comfort equipment, which lead to time consuming diagnosis sessions with traditional computer assisted diagnosis systems mostly relying on expert rules. Therefore, the diagnostic algorithm of such diagnosis systems needs to be improved. In this paper, the authors report their experience of the enhancement of a knowledge-based system to optimize the diagnosis on the basis of a meta-heuristic approach. The experimental platform used for the experimentations is the car diagnostic station SIDIS Enterprise developed by Siemens AG. Based on the integration of an appropriate strategy, the convergence to the faulty component is calculated with accuracy, as well as the lifetime, tests costs, suspicion rules and causal relations. The authors analyzed this methodology for different values of the weight parameters and report the results on a case-based consideration in the automotive domain.