Research On Preventive Maintenance Decision-making Model And Application Based On Delay Time Theory

The production equipment maintenance management is the important factor of manufacturing enterprises production. If the failure happens, it can cause huge economic losses to the enterprise; so the reliability of production equipment is more important. Traditional equipment maintenance management theory and method have many problems. The condition-based maintenance strategy is based on state of system. This strategy is very good to solve the shortage of the traditional maintenance mode. Based on the delay time theory and the practical preventive maintenance (PM) strategy in the enterprise analyze to solve practical problems. In order to reduce the incidence of equipment failure and maintenance cost/total downtime, the paper spreads out from four following respects.The main research contents and innovation points are as follows:(1) A two-phase inspection schedule and an age-based preventive replacement (PR) policy for a single plant item are proposed to reduce the expected cost per unit time. The three stages are referred as the normal, minor defective and severe defective stages. Once an inspection identifies that an item is in the minor defective stage, we may delay the PR action if the time to the age-based PR time is less than or equal to a threshold level, otherwise, replace immediately. A bee colony algorithm is developed to find optimal solution for the proposed models. A numerical example showthe model can effectively reduce the expected cost per unit time and simulations were conducted to verify the correctness of the model.(2) In the above research content a bee colony algorithm is used for solving multidimensional decision-making variable PM model. However, the bee colony algorithm is easy to fall into local optimal solution and low accuracy. Sowe research the bionic mechanism of the bee colony algorithm, the structure process of the solution and related set of parameters to improve the optimization performance of the algorithm.The global convergence of simulated annealing algorithm and the local convergence of bee colony algorithmare combined together so that improved bee colony algorithm. A numerical example is proposed and the results show the the improved bee colony algorithm can effectively solve multidimensional decision-making variable PM model.(3) Condition monitoring (CM) and manual inspection are used in industry to identify a system’s state. A model is presented that considers a single-unit system subject to both CM and additional manual inspections. There are two preset control limits:an inspection threshold and a PR threshold. When a CM measurement is equal to or greater than the inspection threshold but is less than the PR threshold, a manual inspection activity is initiated. When a CM measurement is greater than the PR threshold, a PR activity should be carried out. The system’s degradation process evolves according to a two-stage failure process: the normal working stage and the delay-time stage. We assume that a manual inspection is perfect.The decision variables are the CM interval and the inspection threshold, and we aim to minimize the expected cost per unit time. A numerical example demonstrates the applicability and solution procedure of the model.(4) The PM decision-making research of complex systemis considered. Here the oxygen gun system in some steel enterprise is seen as the research object. Through collecting and analyzing original maintenance record, two-phase delay time theory is used to bulid a statistical model. The maximum likelihood function is constructed to estimate the model parameters. Then we consider the impacts of PM cycle on the total shutdown time over a renewal cycle and bulidPM model. By optimizing the model, the optimal PM cycle is obtained. At modeling time, all possible defects in accordance with the risk degree and the maintenance time length is divided into two categories:big defects and small defects. By using Akaike Information Criterion (AIC), we choose the appropriate forms of distribution, and then by chi-square goodness of fit test, we verify the above options. The model optimizaiotn results show that the steel enterprise to adopt the actual PM strategy does not conform to the optimal PM strategy formulation.