【作者】 周萧；

【导师】 王万良； 杨旭华；

【作者基本信息】 浙江工业大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 随着经济规模的不断扩大,生产规模越来越大,复杂性也越来越高,对企业的管理和大规模的生产过程的监控都提出了更高的要求。对生产调度问题的研究已有十几年的历史,提出了大批的调度方法,但是仍没有形成一套系统的理论与方法。差分进化算法(Differential evolution algorithm, DE)作为一种新颖的智能算法,首先由Storn K和Price P在1995年提出。后经过实践证明,差分进化算法是解决多项式问题的有效方法。但差分进化算法本身存在缺陷,如容易出现早熟收敛,陷入局部最优。本文对基本差分进化算法进行了一些改进,以便更好的解决调度问题。本文的研究内容主要包括:(1)差分进化算法的改进。提出一种混合差分进化算法(Hybrid Differential Evolution Algorithm, HDE)。通过标准benchmark问题仿真结果分析,改进后的算法收敛速度快,全局收敛能力更强。(2)置换流水车间调度问题。通过标准FSP调度问题和实际化工车间调度问题仿真结果分析,验证了混合差分进化算法在解决FSP单目标和多目标调度问题上的有效性和优越性。(3)作业车间调度问题(Job Shop Scheduling Problem,JSP)。差分进化算法在JSP问题中的应用目前还很少。通过针对JSP单目标和多目标问题的仿真结果分析,验证了HDE在解决JSP问题中的可行性和有效性。(4)基于HDE的离子膜车间调度问题。本文调度方案采用基于统一时间离散化的方法,满足多项约束条件的基础上,建立以产值效益最大化为调度目标的调度模型。通过分析DE和HDE算法仿真结果,验证了HDE在解决实际车间调度问题上的优越性。最后对本文进行总结分析,并对以后的工作进行展望。更多还原

【Abstract】 With the continuous expansion of economic scale, bigger production scale and a higher complexity requirements, putting forward higher requirements on management and monitoring of large-scale production processes. The study of production scheduling problem has underwent more than ten years of history and a large number of scheduling methods has been put forward, but still no systematic theory and method has been formed. Differential evolution algorithm as a novel intelligent algorithm first proposed by Storn K, and Price P made in 1995. Later it’s found that the differential evolution algorithm is an effective way to solve polynomial problems. But the basic differential evolution algorithm itself is flawed, for it’s prone to premature convergence into local optimum. In this paper, the basic differential evolution algorithm made some improvements in order to better solve the scheduling problem. This study should include:(1) An improved of differential evolution algorithm. Proposing a hybrid differential evolution algorithm (Hybrid Differential Evolution Algorithm, HDE). Through the simulation results of standard benchmark problems, the improved algorithm has fast convergence capability and excellent global convergence capability.(2) The permutation of flow shop scheduling problem. Through the simulation results about the standard FSP scheduling problem and the actual chemical scheduling problem, the hybrid differential evolution algorithm is effective and superior in solving the scheduling problem on a single and multiple objectives.(3) Job-shop scheduling problem. At present, differential evolution algorithm has not been widely applied to JSP scheduling problems. Through the simulation results about multi-objective and single- objective JSP scheduling problems, HDE is proved to be feasibility and effectiveness in resolving the JSP scheduling problem.(4) The resolution of lizimo workshop scheduling problem based on HDE. Based on time discretion, this scheduling scheme is a unified approach to meet a number of constraints and to establish a scheduling model with the scheduling objective of the value of maximum benefit. By analyzing the simulation results in using DE and the HDE algorithm, it verifies that the HDE has advantages in settling the practical scheduling problem.Finally, this paper ends with a summary and the work of the future outlook.更多还原