Abstract: Because conceptual design of mechanic product has many fuzzy features, fuzzy comprehensive evaluation becomes an advantageous model. If fuzzy evaluation matrix is obtained, the key step in fuzzy evaluation is to realize membership conversion. However the existing methods are wrong, because the redundant data in index membership that are of no use in scheme classification are used to calculate scheme membership. In order to avoid this kind of interference with membership conversion, the most important thing is to ascertain the classification information from index membership for scheme classification. Thus by finding knowledge information in index membership about scheme classification, data mining based on entropy clarifies the relationship between scheme classification and index membership. Classification weight is defined to eliminate the redundant data and classification information is obtained. Membership conversion is realized accurately. Then fuzzy comprehensive evaluation for conceptual design of mechanical product based on new membership conversion is built.

Key words: Classification weight, Conceptual design, Effective value, Fuzzy evaluation, Membership conversion, cable harness wiring, chaotic mutation, complex mechatronic products, multi-scale, particle swarm optimization