Research GSO Set Orders Scheduling Algorithm

In recent years, more and more manufacturing enterprises convert theirstrategic to customer-centered strategy. They use the production mode of ’maketo orders’ to meet customer needs. During the process of production, the biggestconcern of enterprise’s managers is how to improve customers’ satisfaction andreduce their production costs through the reasonable production scheduling.With intelligent optimization algorithm theories becoming more and moremature, a few optimization algorithm have been applied to solve whole-setorder problem in single-machine and achieved some results,for example geneticalgorithm(GA),particle swarm optimization (PSO). However, with the increaseof the problem scale, some of them are easy to fall into these phenomenons,such as premature convergence, local optimum. As one of intelligentoptimization algorithm, Glowworm swarm optimization(GSO) has thecharacteristics of less adjustable parameters,fast convergence rate. It has beenverified with good performance in combinatorial optimization, productionscheduling and so on. According to the characteristics of GSO and whole-setorder problem, this algorithm can not only improve efficiency in singlemachine,but also provides some thingkings of solving practical problems andrelated studies.In this paper, by analyzing the characteristics of whole-set orders problemand combining the theory of glowworm swarm optimization, a glowwormswarm optimization scheduling algorithm is proposed to solve single schedulingproblem firstly. A new hybrid-encoding schema combining withtwo-dimensional encoding and random-key encoding is given. In order toenhance the capability of optimal searching and speed up the convergence rate,the dynamical changed step strategy is integrated into this algorithm.Furthermore,The algorithm’s good performance in solving single schedulingproblems is verified by the experiment results.Secondly, considering that workpieces in multi-process groups,the setuptime could not be ignored, a hybrid-GSO with improved population strategy and timely crossover operation is proposed. By improving the method of generatinginitial population,the quality of the group’s individuals is mended.Through thepopulation average fitness values to judge whether the crossover operation isneeded.Therefore, this method not only can keep the diversity of the offspring,but also avoid the algorithm falls into local optimum.In view of the diversification needs of customers and companies in actualproduction, a multi-objective model is established and a multi-objective GSO isdesigned to solve this problem. This algorithm use the strategies of random-keyencoding and improved step strategy. Its feasibility and effectiveness arevalideated by the experimental results.