【作者】 李平；

【导师】 夏绪辉； 唐秋华；

【作者基本信息】 武汉科技大学 ， 机械设计及理论， 2013， 博士

【摘要】 在现代制造模式中多品种、小批量生产愈来愈多，对产品成本和质量的要求越来越高，因此对车间运作管理也提出了标准化、精细化的要求，致使管理者愈发关注生产中存在的不确定性及其对生产的影响。在实际工作中信息的获得具有不及时和不完整的特点。生产调度需及时了解、充分考虑这些影响因素，在调度方案制定前需防范此因素对生产造成的不平衡隐患，在执行过程中调度方案需随时动态调整以适应这些变化。在总结以往工作的基础上，本文研究混装和作业车间调度时处理不确定的框架、机制和措施，提出在不确定条件下的鲁棒调度方法和动态自适应反应式策略，并对生产过程中的不确定信息处理和参数校正方法进行探讨。本文的主要工作如下：以系统性消除不确定因素的影响为目标，构造了结合预防式调度、反应式调度与不确定推理于一体的整体调度框架。以具有不确定吸收能力的鲁棒调度方案作为生产开始前的预调度方案，基于调度结果通过贝叶斯推理对预估的不确定参数分布进行再处理和修正；利用具有不确定反馈能力的反应式调度，应对生产中各种突发事件，评估并修正应对策略，为下一阶段的预防式和反应式调度提供更可靠的决策依据。基于预防式调度思想，研究具有不确定吸收能力的鲁棒调度方法。针对具有不确定操作时间的混合装配线平衡问题，基于混合整型线性规划建立相应的鲁棒对等模型；针对具有不确定操作时间的作业车间调度问题，建立基于调度目标期望值的目标规划模型并开发相应的智能算法对模型进行求解。针对生产过程中出现的设备故障、订单改变等突发事件，通过调整系统参数中的设备和工件等，提出具有自适应能力的反应式调度方法。并针对柔性作业车间调度问题，开发具有双层编码的遗传算法。研究不确定信息的处理及不确定参数的校正方法。以随机变量的上界、下界、均值和方差为不确定参数描述手段，建立基于随机变量的鲁棒解与基于均值的确定解之间的对应关系。以贝叶斯网络为工具，结合后验信息与先验统计进行分布参数的校正处理，以获得更符合实际情况的分布参数。为降低调度问题的计算复杂性，研究两种快速算法——针对装配线平衡问题的摹加代数方法和针对作业车间调度的Hopfield-神经网络算法。对于前者，通过数学命题证明在摹加代数意义上，简单装配线平衡问题可等价于旅行商问题；对于后者，基于Lyapunov稳定性理论证明方法的收敛性。并通过实际算例验证两种方法的有效性。更多还原

【Abstract】 In the modern manufacturing mode, there are more and more variety and small batchesproduction,lower cost and higher quality,so standardization and fine are put forward formanufacturing workshop operation management, which cause that managers increasingly focuson existed uncertainty in the production and its impact on the production.The information thatget in the practical work is not timely or incomplete.Because science and technology and marketchange quickly,in order to respond to these challenges,it put forward a higher and morestringent requirements for Workshop’s uncertain scheduling method and technology.Productionscheduling need understand it in a timely manner and taking into full account these factorsfully,this will prevent and eliminate the imbalanced danger in production,the scheduling schemein the execution process needs to dynamically adjust to these changes at any time.Based on the summaries of the past work,this parer puts forward the frame and mechanismfor uncertain dynamic scheduling, analyzes the impact on production process cased by uncertaininformation, proposes robust scheduling method and dynamical adaptive reactive strategy underuncertainty and explores the uncertain information processing and parameter correction methodin the production process.The main work and research results are shown as follows:This paper analyzes dynamic scheduling mechanism under uncertain environment, comesup with the overall scheduling framework combined with the scheduling of proactive andreactive scheduling, set robust scheduling with uncertain absorption capacity as thepre-scheduling scheme before start of the production, take the strategy combined withevent-driven strategy and the receding horizon in production, develop the adaptive responsescheduling algorithm in response to emergencies, and use Bayesian filtering algorithm toreprocess and correct uncertain information to provide a more reliable basis for decision makingfor pre-scheduling in the next phase.Concerning assembly line balancing problem and more universal job shop schedulingproblems,this paper analyzes the robust scheduling method that has uncertain absorption capacity.For the assembly line balancing problem, this paper establishes a robust integer linearprogramming model that can cope with the problems with uncertain operation time parameters.this paper establishes goal programming model based scheduling target expectations and developthe intelligent algorithm for job shop scheduling problem to solve the problem.Concerning the emergencies and disturbances in the production process, this paperinvestigate reactive scheduling method. This paper develop adaptive double-coded geneticalgorithm for flexible job shop scheduling problem.According to the universal computational complexity for the scheduling problem,this paper develops two fast algorithms-G plus algebraic method for assembly line scheduling problem and hopfield neural network algorithm for job shop scheduling problem.For the former,the assembly line balancing problem is equivalent to the traveling salesman problem proved bymathematical proposition in the sense of G plus algebraic. For the latter, Convergence of themethod is proved based on Lyapunov stability theory. And the effectiveness of the two methodsis verified.This part researches on processing method for uncertain information and the correctionmethods for the uncertain parameters.Concerning the uncertain parameters in the productionoperating system, this paper set Bayesian theory as a tool, use production data in the posterior tocorrect Prior statistical distribution parameters in order to obtain a more realistic distribution ofthe parameters and provide a reliable guarantee for accuracy and precision in the followingproduction process.更多还原