Hybrid genetic algorithm to minimize scheduling cost with unequal and job dependent earliness tardiness cost

This article presents two combinatorial genetic algorithms (GA), unequal earliness tardiness-GA (UET-GA) and job-dependent earliness tardiness-GA (JDET-GA) for the single-machine scheduling problem to minimize earliness tardiness (ET) cost. The sequence of jobs produced in basic UET and JDET is added to the random population of GA. The best sequence from each epoch is also injected as a population member in the subsequent epoch of GA. The proposed improvement seeks to achieve convergence in less time to search for an optimal solution. Although the GA has been implemented very successfully on many different types of optimization problems, it has been learnt that the algorithm has a search ability difficulty that makes computations NP-hard for types of optimization problems, such as permutation-based optimization problems. The use of a plain random population initialization results in this flaw. The main objective of the article is to develop the combination of heuristics (UET, JDET) and the GA to obtain convergence by reinforcing the random population initialization and finding a promising solution for the reduction in total ET cost. The cost is further significantly lowered offering the due date as a decision variable with JDET-GA. Multiple tests were run on well-known single-machine benchmark examples to demonstrate the efficacy of the proposed methodology, and the results are displayed by comparing them with the fundamental UET and JDET approaches with a notable improvement in cost reduction.