Research On The Optimization Strategy Of Preventive Maintenance For A Single-unit System Based On Multicriteria Decision-making

Maintenance management of manufacturing systems is a significant part of the daily operation of the manufacturing industry,which can guarantee equipment running effectively and safely.With the development of industrialization,traditional maintenance strategies can not meet the increasing industrial demand.This dissertation studies the condition-based maintenance(CBM)strategy of single-unit systems based on multicriteria decision making to reduce equipment failure occurrence and solve the shortage of the traditional preventive maintenance(PM)model from the following four aspects.The main research contents and innovative achievements are as following:(1)Considering that the degradation process of a single component system follows a two-stage failure process based on the delay time theory,including a normal stage and a delay-time stage.A CBM strategy is proposed via the optimization of an availability-cost hybrid factor and consideration of quality control for a single-unit system during a two-stage failure process.The system state changes from the in-control state to the out-of-control state during the production process.Once it enters the out-of-control state,non-conforming items are produced.Minor inspections are carried out periodically to promptly identify the state of the system.Once an out-of-control state is detected,a minor repair is carried out immediately to bring the system back to its in-control state.A perfect major inspection is performed to determine if the system state is beyond a predetermined PM threshold.The system is renewed either by a major repair or a failure repair.The optimal major inspection interval and minor inspection interval are determined by taking the expected cost per unit time and equipment availability as the optimization objectives.(2)Given the existing joint decision models of PM and spare parts inventory of single-unit systems,single-objective optimization cannot balance different demands of manufacturing industries.Considering the expected cost per unit time,availability,shortage of risk as the utility attributes,a joint model of imperfect PM and spare parts inventory is proposed based on the multi-attribute utility theory.The expected cost per unit time and the long-run availability of the single-unit system are modeled using the renewal reward theorem.The system will be renewed in two cases:preventive replacement and failure replacement.The optimal imperfect PM interval and spare part number is obtained simultaneously through the enumeration method and computer programming.A numerical application is illustrated to compare the results of the single-objective model with the multi-objective model.(3)Considering the measurement error of the observed system condition,a joint optimization model of CBM and economic production batch policy is proposed.The degradation process is described by the stochastic coefficient regression model,which consists of the actual deterioration condition and the random error,where the mean value and variance increase with time.Once the observed condition is equal to or higher than the PM threshold,the system should be renewed by preventive replacement immediately.The system is also renewed once a failure occurs.The expected cost per unit time and availability is taken as utility attributes.The optimal economic production batch and PM threshold values were obtained based on the multi-attribute utility theory.A case study of a steel factory producing steel pipes is given.The result demonstrates that the model based on the multi-attribute utility theory can guarantee availability while minimizing the expected cost per unit time.(4)Considering multi-level thresholds,further research on the joint problem of CBM and economic production batch policy is proposed based on the stochastic coefficient regression model.Three types of actions are performed:Imperfect PM,preventive replacement and corrective maintenance(CM).An availability-cost hybrid factor is proposed to balance the cost per unit time and the system availability.Numerical simulations are provided to illustrate the method.A comparison with single-level threshold strategies demonstrates the effectiveness of the proposed multi-level threshold approach.