Exact methods in disassembly sequencing as a benchmark for heuristic algorithms

摘要

This paper deals with the application of exact methods in disassembly sequencing subjected to sequence dependent costs to the development of heuristics. Application of exact methods is restricted to fairly simple products, because CPU time tends to increase exponentially with product complexity, which makes these methods inappropriate for complex products. One has to rely on heuristic methods that typically require a CPU time that polynomially increases with product complexity. As these methods typically return suboptimal solutions, a reliable benchmark is required that can be found in exact solutions. This application is demonstrated on a complex product with an extensive search space. It is demonstrated that a heuristic can be developed that generates a set of extremely good suboptimal solutions.