基于延迟时间理论的预防维修决策模型及应用研究

发布时间：2018-03-10 09:42

本文选题：预防维修　切入点：延迟时间　出处：《北京科技大学》2016年博士论文　论文类型：学位论文

【摘要】：生产设备维修管理是制造企业生产中一项重要的要素,设备一旦发生故障对企业可能会造成巨大的经济损失,因此生产设备的可靠性越来越重要。传统的设备维修管理理论和方法存在诸多问题。基于状态维修策略是以状态为依据的维修。该策略能够很好的解决传统维修方式的不足。本文基于延迟时间理论,结合企业维修管理中采用的实际预防维修策略,分析和解决企业在制定预防维修策略时存在的实际问题,以降低设备故障发生率和维修成本／总停工时间为目的,展开以下四个方面的研究。主要研究内容和创新点如下：(1)在单一部件系统的预防维修策略研究中,考虑加入两阶段检测策略和基于年龄预防更换策略,目的是为了降低单位时间内期望成本。系统经历三个阶段分别为正常阶段、初始缺陷阶段和严重缺陷阶段。一旦系统在检测点被识别出处于初始缺陷阶段时,如果该检测点距离基于年龄预防更换时刻的时间间隔小于事先设定的某一阈值时则考虑对系统实施延迟预防维修策略,否则,系统立即被更新。使用蜜蜂群优化算法求解模型获得最优预防维修策略。数值算例表明该模型能够有效的降低单位时间内期望成本,仿真算法被提出充分证明了模型构建的正确性。(2)上述研究内容中使用蜜蜂群优化算法求解多维决策变量预防维修模型。然而蜜蜂群优化算法易陷入局部最优解和精度低,因此本论文研究蜜蜂群优化算法仿生机理,解的构造过程及相关参数设置,从而进一步改进算法优化性能。将具有全局收敛性的模拟退火算法与具有局部收敛性的蜜蜂群算法结合,提出改进的蜜蜂群优化算法。通过数值算例验证了改进的蜜蜂群优化算法能够有效的求解多维决策变量预防维修模型。(3)在企业预防维修行为中在线状态监测和人工检测被实施是为了识别系统的状态。考虑对单一部件系统联合实施在线状态监测和人工检测预防维修行为。事先设置两个控制线：检测阈值和预防更换阈值。当在线状态监测测量值处于检测阈值和预防更换阈值之间时,实施人工检测。一旦确认系统处于延迟时间阶段则更新系统,否则不采取任何维修措施。当在线状态监测测量值大于预防更换阈值时,更新系统。系统经历两个阶段分别为正常阶段和延迟时间阶段。假设人工检测是完美的。状态监测间隔和检测阈值是决策变量,目的是为了最小化单位时间内期望成本。数值算例表明所提出的预防维修模型是有效的。仿真算法充分证明了模型构建的正确性。(4)考虑复杂系统的预防维修决策研究。以某钢厂企业中的氧枪系统为研究对象。通过搜集和分析原始维修记录,使用两阶段延迟时间理论构建统计模型。采用最大似然方法对模型参数进行参数估计,然后考虑预防维修周期对总停工时间的影响建立预防维修模型,通过优化该模型得到氧枪系统的最优预防维修周期。建模时将所有可能出现的缺陷按照危险程度和维修时间长短分为两类：大缺陷和小缺陷。通过赤池信息准则选择合适的分布形式,然后在通过卡方拟合优度检验方法对其进行验证。模型优化结果表明：该钢厂企业制定的实际预防维修策略不符合最优预防维修策略的制定。
[Abstract]:Maintenance management of equipment manufacturing is an important factor in the production of enterprises, equipment failures of enterprises may cause huge economic losses, so the reliability is more and more important. The traditional production equipment maintenance management theory and method has many problems. Based on the condition based maintenance strategy is based on the state of maintenance. The strategy can solve the shortcomings of traditional maintenance mode. This paper based on the theory of time delay, combined with the maintenance and management of the enterprise actual preventive maintenance strategy, to analyze and solve practical problems in the development of preventive maintenance strategy of enterprises, in order to reduce the equipment failure rate and maintenance cost / total downtime for the purpose of the following in the following four aspects. The main research contents and innovations are as follows: (1) in the study of preventive maintenance strategy for single components of the system, consider joining the two stage The detection strategy based on age and preventive replacement policy, the purpose is to reduce the expected cost per unit time. The system has experienced three stages: normal stage, initial stage and defect serious defect stage. Once the system is identified in the detection point comes from the initial defect stage, if the detection point from a threshold age preventive replacement time the time interval is less than the pre-set time delay based on considering the implementation of preventive maintenance strategy of the system, otherwise, the system is updated immediately. The use of bee swarm optimization algorithm to solve the model to obtain the optimal preventive maintenance strategy. Numerical examples show that the model can effectively reduce the expected cost per unit time, the simulation algorithm is proposed to fully demonstrate the correctness of model. (2) the bee swarm optimization algorithm for decision variable preventive maintenance model using the above research content. However, the bees. The algorithm is easy to fall into local optimal solution and low precision, so this thesis bee swarm optimization algorithm of bionic mechanism, set up construction process solutions and related parameters, so as to further improve the optimization performance of the algorithm. Combined with the simulated annealing algorithm has global convergence and local convergence has bee swarm algorithm, an improved bee colony optimization the algorithm is verified through a numerical example. The improved bee swarm optimization algorithm can solve the multidimensional model of preventive maintenance decision variables effectively. (3) in the prevention of enterprise online monitoring and maintenance manual detection behavior is carried out to identify the state of the system. Considering the joint implementation of preventive maintenance behavior of online monitoring and artificial detection of single component set in advance system. Two control line: the detection threshold and preventive replacement threshold. When the measured value in on-line monitoring the detection threshold and prevent more Change the threshold when the implementation of artificial detection. Once confirmed in delay time stage, update the system, otherwise do not take any maintenance measures. When updating the system of on-line monitoring the measured value is greater than the preventive replacement threshold,. The system has two stages respectively in normal phase and time delay phase. Under the assumption that the artificial detection is the perfect state. The monitoring interval and the detection threshold is the decision variables, objective is to minimize the expected cost per unit time. Numerical examples show that the proposed preventive maintenance model is effective. The simulation algorithm fully proved the correctness of the model. (4) the research of preventive maintenance decision making complex system into account. In a steel enterprise in oxygen lance system as the research object. Through the collection and analysis of original maintenance records, the use of two phase delay time theory to construct statistical models. Using the maximum likelihood method to model parameters. For parameter estimation, and then consider the preventive maintenance period of the establishment of preventive maintenance model for the effect of total downtime, the optimal preventive maintenance period of the model of oxygen lance system. When modeling all possible defects according to the degree of risk and the maintenance time is divided into two categories: large defects and small defects. Through Akaike information criteria for selection of appropriate forms of distribution, and then through the chi square goodness of fit test method to verify the model. The optimization results show that the actual preventive maintenance strategy of the steel enterprises set not in line with the development of optimal maintenance strategy.

【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TB49

【相似文献】