Product/Process Configuration Evolutionary Optimization: A Multiobjective Clustering in Order to Reduce Inconsistencies During Crossover

Concurrent configuration of a product and its associated production process is a challenging problem in customer/supplier relations dealing with configurable products. Search for optimized solutions that respect customer's needs and constraints of the problem in a multiobjective context is a particularly difficult task. Constraints Filtering Based Evolutionary Algorithm (CFB-EA) [1] proposes an original way to integrate constraints satisfaction in optimization thanks to a constraints filtering engine. CFB-EA tries to mix solutions randomly selected in order to improve them but leads to many incompatibility occurrences which are time consuming. We propose in this article a dedicated multiobjective clustering algorithm that reduces incompatibilities occurrences and improve the selection of solutions for crossover operator.