Discrete Manufacturing Process Design Optimization Using Generalized Hill Climbing Algorithms

摘要

Discrete manufacturing process design optimization (DMPDO) is a problem of significant importance and interest to the Air Force. Moreover, the complexity of parts that must be manufactured for airplane engines and other units also makes this problem extremely difficult, due in part to the large number of design sequences that exist for each given part. This has forced researchers to develop heuristic strategies to address such design problems. This report summarizes the research that has been conducted in developing the generalized hill climbing (GHC) algorithm framework for discrete optimization problems in general, and the DMPDO problems in particular. New necessary and sufficient convergence conditions for GHC algorithms have been developed that use a new algorithm iteration classification scheme. Moreover, new performance measures are formulated that more closely capture how practitioners use GHC algorithms to solve large-scale real-world problems. These results provide fundamental insights into how such algorithms should be applied, as well as provide answers to open questions concerning the convergence and performance of various GHC algorithms. On-going interactions with researchers at the Materials Process Design Branch of the AFRL at WPAFB and at Austral Engineering and Software, Inc. has resulted in new software modules for GHC algorithms that are being used by to solve various DMPDO problems of interest to the Air Force.