Research On Intelligent Algorithm And Its Application To Production Scheduling Problem Under Uncertainty

Production scheduling, an important part of production planning, is the core and key technology of modern production management, which means that an effective scheduling policy is worked out based on limited resources to optimize economic or systems performance criterion. Therefore, appropriate scheduling policy can not only improve the comprehensive management level, but also make a fast and remarkable economic profit. Also, scheduling problem, in which multi-constraints, multi-objectives, various uncertainties etc. are included, has been proven to be the NP-hard problems.In this paper, two main jobs are completed in great details. For one thing, two kinds of modified co-operative co-evolution particle swarm optimizer (CPSO) are put forward including co-evolution particle swarm optimizer based on niche sharing scheme (NCPSO) and co-evolution quantum-behaved particle swarm optimizer combined with simulated annealing mechanism (SACQPSO), of which the former one strengthens diversity of population, improves the quality of individuals and enhance the convergence performance of the algorithm and the latter one not only enhance the capacity of searching the best solution and increase the diversity of particles owing to co-operative co-evolution thought and quantum-behaved theory, but also strengthen the ability of global searching as the result of simulated annealing. For the other thing, discrete scheduling problems under certainties and uncertainties are depicted systemically, of which the production scheduling models are established respectively. Eventually, the proposed novel algorithms are adopted to optimize some typical production scheduling problems. The main contributions of this dissertation can be summarized as follows:(1)The description and principle of production scheduling problem are summarized comprehensively, which include the feature, classification, current research situation, developing orientation and research method of the scheduling problem. Also, the classification, mathematic description method, research situation and scheduling scheme of the production scheduling problem under uncertainty are presented systemically.(2)Original particle swarm optimizer (PSO) and co-evolution theory are introduced systemically, in which the key opinions made up of the principle, flowing chart, characteristic of PSO and the operation process of cooperative co-evolution method are stated in details. Besides, the principle of the niche sharing scheme, the simulated annealing mechanism, the quantum-behaved theory and self-adaptive variance scheme are described comprehensively.(3)Two novel modified co-evolution particle swarm optimizers containing NCPSO and SACQPSO are applied to settle flow shop scheduling problem (FSSP), job shop scheduling problem (JSSP) under certainties and benchmark testing functions. According the experimental results, the modified algorithms are feasible and more effective than basic co-evolution particle swarm optimizer.(4)Taking uncertain processing time for instance, rough sets theory was employed to transform the rough Flow shop scheduling model into the precise scheduling model. Finally, co-evolution particle swarm optimizer based on niche sharing scheme (NCPSO), characterized by strengthening ability of reserve excellent particles and searching the optimal solution, is adopted to minimize the makespan and experimental results show that the novel algorithm is effective and efficient. Also, flow shop earliness/tardiness scheduling problem with distinct due date and uncertain processing time is discussed, in which triangular fuzzy number is used to denote uncertain processing time. Establishing fuzzy programming model based on E/T scheduling model, the key of the problem, is completed through two fuzzy operators. Then using the algorithm of maximizing the membership function of middle value, fuzzy programming model is transformed into precise single-object programming model. Finally, co-evolution quantum-behaved particle swarm optimizer combined with simulated annealing mechanism (SACQPSO) is employed to optimize the fuzzy scheduling problem. And the experiment results indicate that the novel algorithm is effect and efficient.