An integrated approach of evolutionary algorithms with artificial bee colony algorithm for job shop scheduling problems

There have been a lot of research made about solution of scheduling problems that have a very important place in many areas of our lives for years. The cause of these researches is to develop better than the current schedule and achieve greater profits. Therefore, there is great importance of efficient scheduling for both humans and businesses. In this context, heuristic algorithms are used extensively by researchers for solving scheduling problems in recent years. In this dissertation study, an integrated approach has been developed for optimizing the solution of job shop scheduling problems. In this context, artificial bee colony algorithm and evolutionary algorithms are used for the integrated approach. The proposed hybrid method has been applied to data sets related to job shop scheduling. The obtained results have been compared with the results of different optimization techniques that these techniques are ant colony optimization (ACO), particle swarm optimization (PSO) and differential evolution algorithm (DE) using the average relative error percentage (ARPE) and average relative percentage deviation (ARPD) criteria. It has investigated whether statistically significant differences among methods using parametric and non-parametric tests with the founded hypotheses for the comparisons. According to the ARPE criterion, statistically significant differences have been obtained between the results of the recommended approach and ACO technique. According to the same criterion, statistically significant differences have not been observed between the result of the proposed method with PSO and DE algorithms. ARPE value of the recommended approach yielded 4.3 points (as percentage changes) more effective than ARPE value of the ACO technique according to the results of the tests. According to the ARPD criterion, statistically significant differences have been obtained between the results of the recommended approach and other all techniques. According to the results of the tests, ARPD value of the proposed method yielded more effective and stable of 6.3 points than ARPE value of the ACO technique, of 0.6 points than ARPE value of the PSO algorithm, of 0.7 points than ARPE value of the DE algorithm. According to the results of the tests, it observed that the proposed method has much faster and more effective results in conditions less than 20 number of machines or jobs which will be scheduling.