Research on manufacturing process scheduling problem is a front research direction in advanced manufacturing and automation fields. Facing the practical manufacturing process, this dissertation studies the genetic algorithms for the complex Job shop scheduling problem with the support of the National Basic Research Program of China (973 Program) and the National Natural Science Foundation of China. This dissertation includes the following main contents. 1)In order to overcome the defect of the GA with operation-based representation for solving larger scale Job shop scheduling problem, we propose a self-adaptive hybrid genetic algorithm based on the problem decomposition. On the basis of the defined scheduling characteristic—resource conflicting degree, all operations of the scheduling problem are dynamically divided into two parts, each of which is scheduled by means of different scheduling policies. Additionally, a fuzzy logic controller (FLC) is constructed to adjust adaptively the length of the chromosome so that the search space of the proposed GA could be reduced and the performance of the GA could be improved. 2)For larger scale Job shop scheduling problem with partial due date, we propose a new GA based on the problem characteristic, in which the characteristic of the time constraints is adopted in the crossover and mutation operations. 3)For the large scale Job shop scheduling problem, we propose a kind of self-adaptive decomposition algorithm based on prediction. The scheduling problem is dynamically decomposed into some optimization sub-problems, which are solved by means of the proposed GA. In the proposed algorithm, we determine the optimization sub-problems and decode the chromosomes based on the prediction information, respectively. 4)For the Job shop scheduling problem with fuzzy processing time, we propose a kind of hybrid genetic algorithm based on the approximation of the fuzzy number. First, we use a kind of regular piecewise trapezoid fuzzy number to approximate the irregular piecewise linear fuzzy number for formulating the uncertain processing time of the job, and give the approximation distance under certain conditions. Then, based on the above approximation, we give a new kind of fuzzy number integration method including the addition, the max and the comparison methods of the irregular piecewise linear fuzzy numbers. Additionally, we analyze the computation errors and computation complexity of the proposed addition and max methods, respectively. The numerical computational results reveal that the above algorithms are effective. Additionally, we validate the above partial algorithms using the data from the practical automobile manufacturing process.
修改评论