Research On Module Partition Method Of CoPS

Along with global economic integration and social informization course, the market circumstances presented before the enterprise varies greatly; technical competition of the product become more intensive, the research and development costs are correspondingly elevated; the customer become more interested in personalized and custom-made products. While modularization is an effective way for reducing research and development cost and meeting custom-made requirements, modularization design can not only satisfy the custom-made requirements, but can also shorten the development time and cut down the development cost. And it can be good basis for the innovation. Therefore modern modularization theory emphasizes more on complex product system; complex product system generally refers to large products and systems which are characterized by high development cost, intensive technology, complex structure, high degree of custom-made requirements. It is closely related to modern industrial system, and is concerned with national economic arteries. Thus it can be concluded that the research of complex product systems modularization of both theoretical and practical significance.In modern modularization design process, module partition(or modular ecomposition) is an very important part and one of the key technologies, rationality of module partition is related to the "modularization" successiveness of a product system. In the modularization process of a complex product, the integrity of the product system should be considered as well as the correlation of its internal components, in the same time other factors of the life cycle period should also be considered synthetically to achieve a more rational module partition result. How to realize this target becomes the key of the research, therefore module partition method used in complex product system is presented in this paper, concrete research contents should include the following aspects:On the basis of constructing a module partition hierarchical model of complex product system, module partition principles for complex product system is presented. Through analysis of the research status of modern modularization design theory, according to the characteristics of complex product system, modularization design of complex product system is treated by introducing conceptual design ideology, hierarchical model of the partition based on function-principle(technology)-behavior-structure(F-P(T)-B-S) mapping process, and module type is defined considering characteristics of complex product system, still module partition rules for complex product system is presented based on the module, that is, function, principle technology, behavior, structure and physical correlation of its internal components should be considered synthetically in module partition process, which can be referred as partition principle based on F-P(T)-B-S; and other influencing factors should be further considered in a whole view, which can be called partition principle of X design oriented (DFX), in which X means various stages of the life cycle period of the system and factors that affect the competitiveness(such as time, cost, etc.).As to module partition principles for complex product system, various analysis approaches are presented to implement different granularity partitions for the system. Through analysis of the customer requirements and function structure of the system, according to module partition principles, hierarchical fuzzy cluster method is used first to get coarse partitions, in which some of them are picked up as rational partition method employing evaluation method;and number of the modules and important components contained in every module are determined, which function as the initialization parameters for fine module partition, In order to eliminate the defects of local optimum which is incurred by initialization sensitive characteristic of PSO and progressive fuzzy cluster method, an improved combined scheme of the two methods is presented. Finally fine partition based on improved particle swarm optimization based on weighted FCM is introduced, which is the optimizing process of coarse partition that leads to an optimal module partition scheme.Coarse module partition for complex product system will result in multiple partition schemes. The scheme is evaluated or favored employing multi-level gray fuzzy evaluation method after setting up index model of multi-target evaluation.The scheme set is built according to CoPS coarse module partition scheme; Then multi-level evaluation index model is built from the angle of DFX, and this model can be used to evaluate the partition scheme comprehensively; after that ANP method is presented to achieve the subjective weight of the index considering the characteristic of the evaluation index model, and weights of various level indexes are given when objective weights deduced using entropy method are combined with subjective weight, which means ANP-E weighting method of subjective & objective combination. Finally, as to fuzzy mathematics and gray theory, the module partition scheme of complex product is evaluated or favored employing comprehensive multi-level gray fuzzy evaluation method.Here mechanical close experimental platform system of marine bevel gearboxes is used as an example to verify the module partition method for complex product system presented in this paper. First, through function structure analysis of experimental platform system, it is simplified on high level structure composition of the system, and module partition hierarchical model of F-P(T)-B-S the function- principle technology- behavior-structure is built up. Then from angles of function, principle technology, behavior, structure and physical correlation, its internal component correlations are analyzed. Finally improved particle swarm optimization algorithm and fuzzy clustering analysis method are used together to get module partitions ranging from coarse to fine partitions, and various schemes arise from coarse partitions are evaluated or favored employing multi-level gray fuzzy comprehensive evaluation method after setting up index model of multi-target evaluation. The example proves that the approach can be used to determine more precisely which module one component belongs to and finish module partition. The partition result is in correspondence with the function analysis and practical design, and for complex product system, the availability and applicability of the module partition method are thus testified.