Model-Based Decomposition Using Non-Binary Dependency Analysis and Heuristic Partitioning Analysis

The two-phase method for model-based decomposition (Chen et al. 2005a) has two major functional components: dependency analysis and partitioning analysis. The functions of these two components are enhanced and generalized in this paper in order to improve the method’s capability. On the one hand, the non-binary dependency analysis is developed such that the two-phase method can handle both binary and non-binary dependency information of the model. The essence of this development is to properly select a resemblance coefficient for the quantification of couplings among the model’s elements. On the other hand, as the past version of partitioning analysis takes the enumerative approach to search decomposition solutions, the heuristic partitioning analysis is developed as an alterative to search a reasonably good solution in a shorter time. The working principle of the heuristic approach is to analyze the coupling structure of the model such that the weak coupling links among the model’s elements can be identified for model partitioning. At the end, a relief valve system is applied to illustrate and justify the newly developed method components.