A Multi-level Hybrid Framework for the General Flow-shop Scheduling Problem

Despite the large amount of work performed on flow-shop scheduling most of the research has concentrated on the permutation problem in which passing is not allowed. In this work the general flow-shop problem, in which passing is allowed, is dealt with as it is considered to be a better representation of flowshop instances. The evolutionary techniques of scatter search (SS) and its generalised form, path relinking (PR) are applied to these problems as they are able to provide a wide exploration of the search space and they can be integrated with intelligent search methods such as tabu search. The SS and PR strategies are embedded within a core and shell framework. Initiated from a powerful starting solution the core consists of a highly constrained neighbourhood, estimation strategies and a dynamic tabu tenure which provide efficiency and effectiveness and maximise the performance of the core within the various improving and disimproving phases of the search. Several shell strategies, applying specific and generic sequence information, are superimposed on to the core in order to provide the necessary mixture of intensification and diversification. This framework is able to provide substantially better results than the tabu search approach of Nowicki and Smutnicki (1996b). Comparisons on the available benchmark problems indicate that the average deviation from optimum is 8.475 % with 52 best solutions equalled and 41 new best solutions achieved. page (2) A multi-level hybrid framework for the general flow-shop scheduling problem 03/10/98