The Research On Multi-objective Production Scheduling Based On Genetic Algorithm

The production process of clothing enterprises is very complicated and labor-intensive, the categories of products are various. As the change of market demand, orders become more diverse, urgent and random. The clothing enterprises basically rely on the experienced management skills to make production scheduling manually, it is hard to integrate the resource constraints and objects together and ensure the balance of production capacity. That is not scientific and reasonable. So clothing enterprises have to come up with a scientific and rational way to improve economic efficiency.As there are many clothing enterprises, this paper takes a clothing enterprise in Jiangyin of Jiangsu province as an example. It has two different kinds of orders and scheduling lines and two scheduling goals are to minimize the maximum completion time and minimize the maximum tardiness. The orders that have small quantity are scheduled on single scheduling line; the orders that have big quantity are scheduled on mixed scheduling line. However, the artificial experience method have problems such as not efficient, poor balance, easy to delay the delivery date and so on.Aim at these problems, this paper simplify the single machine scheduling problem to permutation flow shop problem and establish a multi-objects single scheduling model. The model overall considerate resource constraint, the operation time and sequence constraint. Then, these papers simplify the mixed machine scheduling problem to hybrid flow shop problem. As workers’ proficiency is not the same, the processing proficiency on different machines is also different. So this paper add the works’proficiency factor, this reflect on the different production efficiency of different machines in the same process. And establish the multi-objects mixed scheduling model. When solve the multi-objective problem via genetic algorithm, this paper come up with a selective dynamically adjusted weight method to design the weights of different objects. This method measures the improvement degree for each generation, Using it to calculate the weight coefficient of each generation population to obtain the selection pressure on improving direction, design the fitness function. So translate the multi-objects problems to single object problem. Then, this paper design the real number encoding method base on artifact and the matrix real number encoding method to solve single flow shop problem and hybrid flow shop problem separately. At last, compare with the result of artificial experience scheduling method using the production examples, it not only demonstrates the improvement of scheduling effectiveness, but also reduce the processing time and tardiness, improve the production efficiency of enterprises.