Dynamic Adaptive Weighted Polymorphic Ant Colony Algorithm For Scheduling Batch-processing Machine With Non-identical Job Sizes

The scheduling of batch-processing machine with non-identical job sizes has been widely used in modern industrial production. Research of the related scheduling algorithm proves to be helpful in achieving higher productivity and market competitiveness of the manufacturer. This paper is aimed to present a deeper study of dynamic adaptive weighted polymorphic ant colony algorithm, which is usually applied to both a single batch-processing machine and parallel batch-processing machine with non-identical job sizes. With that, the study goes on to show how research of batch-processing machine with non-identical job sizes is extended from the deterministic manufacturing environment to fuzzy manufacturing conditions, which is expected to make an easier application of the algorithm to actual production.To begin with, the dynamic adaptive weighted polymorphic ant colony algorithm is applied to minimize the makespan on a single batch-processing machine with non-identical job sizes. The algorithm benefits from the idea that each of different types of ant colonies has different assignments and an information-updating mechanism. At the same time, in solving the problem that polymorphic ant colony algorithm is only too easy to come into the local optimum easily, or cannot carry out all jobs directly, some improvements are made:introduction of weighting factors in the transition probability to prevent repeated search of one job with others abandoned in the basic polymorphic ant colony algorithm; dynamic adaptation of the pheromone update mechanism of the search ants by adjusting the volatile pheromone and pheromone intensity factor, with the combination of the global updating strategy of the basic ant colony algorithm and the local optimal updating strategy of the polymorphic ant colony algorithm, with the aim of circumventing incidental non-convergence of the basic ant colony algorithm and quick occurrence of local optimum in the polymorphic ant colony algorithm. A number of simulations have showed much more improvement being done in dynamic adaptive weighted polymorphic ant colony algorithm than in others. Second, this paper provides a detailed analysis of the effects of both individual parameters and their combinations in the algorithm upon the performance of the algorithm. For this purpose, orthogonal experiments are made twice for selecting parameters of the dynamic adaptive weighted polymorphic ant colony algorithm for a single batch-processing machine with non-identical job sizes, and the best combination is finally obtained.Then, the dynamic adaptive weighted polymorphic ant colony algorithm is applied to a single batch-processing machine with non-identical job sizes in fuzzy manufacturing environment, and parallel batch-processing machine with non-identical job sizes both in the deterministic and the fuzzy manufacturing environment. A series of computer simulations of different levels of instances have proved the effectiveness and applicability of the dynamic adaptive weighted polymorphic ant colony algorithm in handling those mentioned scheduling problems.