Hybrid evolutionary techniques for the maintenance scheduling problem

Abstract

The incorporation of local search operators into a genetic algorithm has provided very good results in certain scheduling problems. The resulting algorithm from this hybrid approach has been termed a Memetic Algorithm. This paper investigates the use of a memetic algorithm for the thermal generator maintenance scheduling problem. The local search operators alone have been found (in earlier work by the authors and others) to produce good quality results. The main purpose of this paper is to discover whether a memetic approach can produce better results. We describe the approach taken and highlight the variety of local search algorithms that were employed. We compare the memetic algorithms with a variety of algorithms that include the local search operators on their own and a range of algorithms that apply the local search operator to randomly generated solutions. We see that, for the problems tested, the memetic algorithms produce better quality solutions (although they do take more time about it). Of course, in practice, for a problem like this, the time taken to produce a solution is not a major issue. What is far more important is the quality of the solution. We conclude that the most effective method (of the ones tested here) is a memetic approach that employs a tabu-search operator.

Hybrid Evolutionary Algorithms: Methodologies, Architectures, and Reviews

“…Memetic Algorithms have been proved to be orders of magnitude faster and more accurate than evolutionary algorithms for different classes of problems. As reported in the literature, hybrid methods combining probabilistic methods and deterministic methods have found success in solving complex optimization problems [4][5][6].…”

Maintenance Scheduling: Description, Status and Future Directions

“…Also, several effective and efficient techniques have been suggested as alternatives to traditional optimisation-based techniques where the problem size is large and the exact optimal solution is not available. The modern techniques include simulated annealing [5,47,71], genetic algorithms [20,21,56,78,85], heuristics [35,51,77], memetics [8], and hybrids [6,49,50]. Various optimisation studies have also been conducted, including by Billinton and colleague [9], Charles-Owaba [13], Christiaanse [15], El-Sharkh [28], Pan [69], Yamayee and associate [86], and Zunn and Quintana [91,92].…”

“…The major efforts, perhaps, are those that have been devoted to studying the traditional optimisation-based techniques such as integer programming [1,2,5,6,22], dynamic programming [5,6,9,10] and branch-and-bound [10,15,16,27] which have been proposed to solve small problems. These methods give an exact optimal solution.…”

A case study of memetic algorithms for constraint optimization

“…In a canonical MA, a prefixed single meme is employed after mutation and evaluation steps of a GA. Obviously, a variety of memes might be designed for solving a specific problem. There is strong empirical evidence that the choice of meme in canonical MAs influences the performance of the search [4,37,38,39,41,55]. Many strategies can be adopted to utilize multiple memes simultaneously within the MA framework.…”