Integrated Modeling And Optimization For Single-Machine Scheduling And Condition-Based Maintenance Decision

Research on integration of production scheduling and equipment maintenance turns out to be a key direction in the field of industrial manufacture for the interactive activities between them. With the development of modern sensor technology, the degradation conditions of an increasing number of equipment can be acquired directly or indirectly. Condition-based maintenance(CBM) policy is a fashion maintenance mode,in which the maintenance activity is arranged according to the detected deterioration state of the equipment. Single-machine scheduling is the basic issue of other complex scheduling, therefore, the research on integrated optimization model for single-machine scheduling and CBM decision becomes a new subject urgently needs to be solved.Focusing on the single-machine production system, this dissertation studies integrated model and optimization for scheduling and CBM decision making. The corresponding work can be summarized in the following:As for the equipment with a known stochastic-distribution failure time, the maintenance strategy combining preventive maintenance based on failure rate threshold and minor repair is developed. In this method, an integrated model is built by minimizing the expected total completion time of jobs as optimization target and solved with genetic algorithm. The results from the comparison between our work and another groups' indicate that the proposed method can further improve production efficiency else.As to the equipment with detectable degradation conditions, a hybrid policy combining imperfect condition-based preventive maintenance and minor repair during job processing as well as failure replacement after processing is presented. Specifically, an integrated model for single-machine scheduling and CBM decision making is established by minimizing the expected total weighted completion time of jobs within scheduling sequence and solved with integration of stochastic simulation, neural network and genetic algorithm. Numerical simulation results show that excessive maintenance or insufficient maintenance during production scheduling process can be avoided effectively.For the high-risk production system in special industry, a condition-based preventive maintenance policy is given and a maintenance decision making model is built. In this case, the performance reliability constraint is considered. Besides, the long-term average cost rate is taken to be minimized as the objective function. With sensitivity analysis, the importance of performance reliability constraint for maintenance decision making outstandings. Based on this, an integration model with performance reliability constraint for single-machine scheduling and CBM decision making is built by minimizing the total expected weighted completion time of jobs as optimization target. According to the hybrid maintenance policy mentioned above, the related calculating general formula of maintenance probability and probability density function can be derived. In addition, the numerical calculation method is elaborated. Similarly, the importance of performance reliability constraint of the integration model is illustrated. In order to verify this model, an application research is conducted on the integration modeling for production scheduling of an electric arc furnace which as a critical equipment in the smelting industry and CBM policy based on degradation condition of the furnace wall.In order to make production efficiency reach the highest level at a certain probability under conditions of satisfied production demands, maintenance demands and reliability constraints, an integrated model is built for single-machine scheduling and CBM decision making subject to stochastic chance constraint programming by taking performance reliability constraint as chance constraint condition, and by minimizing total completion time with given confidence coefficient as optimization target. In addition, several techniques including stochastic simulation, neural network, classifier and genetic algorithm are integrated to a hybrid algorithms for the solution of this model.For the situation where both time of delivery of some jobs and performance reliability being constrained, an integrated model for single-machine scheduling and CBM decision making is built by taking the total processing cost and total completion time as objective respectively. Then the disadvantages of single target integration model are analyzed. After then, a multi-object integration optimization model for single-machine scheduling and CBM decision making is built and solved using multi-object genetic algorithm.Based on failure features and abrasion data of cutters, a cutter degradation model is built. Further, an integrated model having different condition-based preventive maintenance policies is also established for single-machine scheduling and condition-based replacement, depending on the detected abrasion state of knife edge coming from online or offline fashion. Finally, the tap used in machine processing industry is taken as the object for application study on single-machine scheduling and maintenance decision making subject to tap-based cutting teeth abrasion condition.