Research On Multi - Task Scheduling Strategy Based On Improved Genetic Algorithm For 3D Printing

As a new technology,3D printing is one of the key points steering "the 3rd Industrial Revolution". With the rapid development of 3D printing, reasonable and effective scheduling strategy should be pointed out quickly to help companies to achieve the extensive production and win the competition. Based on current researches on scheduling and multi-objective optimization, this dissertation studies multiple tasks scheduling for 3D printing with genetic algorithm (GA) and non-dominated sorting genetic algorithm with elitism approach (NSGA-Ⅱ).The main work of this dissertation is summarized as following:Firstly, as companies’ most concerns are time and cost, this paper proposes three models:mean service time (duration), mean cost, time and cost combined target. Moreover, price difference of printing precision is added in the cost target, which is advanced research for marketing. To improve customers’ satisfaction, two more targets are proposed:waiting time (resources) and deviation of printing precision (quality), where are integrated considering requirements of customers and companies.Secondly, by using genetic algorithm (GA), duration, cost, duration and cost combined target are tackled respectively. Analysis of the revolution process in stage proves convergence of GA in optimizing multiple tasks scheduling for 3D printing. Through single target optimization and comparison mutually, GA is proved to be well efficient. Also, it’s found that GA is easy to prone to premature convergence in the process of optimization with a low computing race and hard to reach global optimum.Thirdly, to avoid premature convergence, the multiple tasks scheduling problems for 3D printers are tackled by using NSGA-Ⅱ. Through combining targets of duration, cost, quality and resource in an organized way, series of non-dominated solutions are got by optimizing the multi-objective problems with two targets, three targets and five targets, which proves the efficiency of NSGA-Ⅱin optimizing multiple tasks scheduling for 3D printing. At last, comparison of NSGA-Ⅱ optimized non-dominated solutions and GA optimized solutions proves that using NSGA-Ⅱ to solve the multiple tasks scheduling problems for 3D printing is more efficient than using traditional GA.