Abstract: A new approach to mechanical kinematic scheme generation based on multi-granularity co-evolutionary function reasoning is proposed. Firstly, features of co-evolutionary function reasoning model and multi-granularity design model are investigated, then a multi-granularity co-evolutionary function reasoning model is constructed by integrating their strongpoint. Such a model is applied into the kinematic scheme generation and a co-evolutionary algorithm for kinematic scheme genera-tion is presented, in which classified functions are used to de-scribe the kinematic characters of mechanical system and mechanisms, each mechanism and mechanical system is ex-pressed by a motion-transforming function abstractly and series-connected mechanical kinematic schemes are generated based on function reasoning. Next, detailed flow of function reasoning algorithm is provided and the characteristics of the algorithm are analyzed by comparing with those of existed algorithms. Further, the advantages of the algorithm, such as high computing efficiency, high precision in describing motion transformation, are discussed. Finally, kinematic scheme gen-eration of a wire-feeding mechanism is used to verify the effec-tiveness of the new approach.

Key words: Co-evolution, Function reasoning, Kinematic scheme generation, Multi-granularity, Crack propagation, Reconstruction, Rolling contact fatigue crack, X-ray Computed Tomography Scan, Rail