| The production schedule and manufacturing process are joined by job-shop scheduling, which is the hinge to efficiency. Supported by National Natural Science Foundation (Grant No.50975039), this thesis studied algorithms to complex Job-shop Scheduling Problem (JSP), which regarded AGV constrained JSP appeared in the company G in Fujian as the engineering background. In practical manufacturing process, dealing with JSP and AGV transportation properly may lower the cost and enhance productivity. How to schedule the AGV and jobs in the workshop in order to coordinate the logistic system and manufacturing system is the crux to boosting competitiveness of an enterprise. Main work of this thesis is summarized as follows:(1) The classic coding schemes were analyzed in the aspects of Lamarckian, completeness and decoding complexity. A relative position coding strategy was proposed based on the ordering nature of the JSP. The hybridized coding schemes were proved to reducing dimensions and the results invalid or reduplicated by experiments.(2) A dual-system optimization algorithm was presented to cope with the JSP, which is NP-hard. The dual-system was based on the "co-evolution" concept, and ran the differential evolution algorithm and genetic algorithm pseudo-concurrently. With the help of elite migration, the population diversity was improved. The results of the experiments showed that the dual-system algorithm could acquire a better solution with inhibiting premature.(3) Based on the study above, a dual-system algorithm with hybridized coding schemes was put forward to solve the AGV constrained JSP. A job-shop scheduling software with an interface to AGV scheduling system was developed. The structure of the software, the database and the user interface were drafted. A simple example proved the platform worked well.The thesis was expected to be helpful to the study of the dual-system and JSP in theory, and the constrained JSP in practical engineering. |
PDF Full Text Request