Research And Application On Rescheduling In Assembly Process Of Steam Turbine

Steam turbine products are of complex structure,huge volume and weighing tens of tons.The price can reach as high as hundreds of millions of yuan.Products are customized according to specific needs of different customers.In order to respond to changes of market demand quickly,the engineering-to-order(ETO)model is adopted.In the actual assembly process,late delivery of materials,abnormal assembly quality and workers operation delay will seriously disturb the initial schedule.It will cause significant economic losses once product delivery is delayed.Therefore,it is necessary to study the rescheduling method in assembly process so that under disruptions assembly operation of steam turbines can be effectively adjusted and controlled,efficiency can be improved,and timely delivery of steam turbines can be ensured.Main work and research results of this paper are as follows:To solve the rescheduling problem of steam turbine assembly process,a rescheduling model aiming at cost minimizing and plan stability was constructed,and a multi-objective algorithm based on imperial competition mechanism was designed.Fast non-dominant sorting mechanism was adopted and initial colonial allocation,assimilation,competition and extinction mechanism were optimized in the algorithm.It also added a colonial reform operation to increase global search ability.Based on the MRCPSP problem set in PSPLIB,parameters of the algorithm were adjusted and performance of the algorithm was tested.Experiments showed that optimal coverage of MOICA was 5% and 17% higher than SPEA and NSGA-II respectively.Compared with the result relied on manual experience,this rescheduling method improved plan stability by 30% and reduced cost by 12% in four abnormal events in the assembly system of K156 cylinder.The problem of cooperative rescheduling between assembly operation and crane logistics was further studied.The cooperative rescheduling model was constructed,two sets of heuristic rules of pre-assignment and immediate assignment were designed,and conflict determination and avoidance rules for double-deck cranes were presented.A multi-objective evolutionary algorithm was proposed.The case of MISTA 2013 competition was used to generate testing data.Testing results showed the pre-assignment rule overwhelmed immediate-assignment rule,and comparative test of two algorithms proposed in this paper was carried out.The method was applied in the actual case of cylinder assembly site and reduced cost by 9%.Finally,the process of plan management in steam turbine assembly site was analyzed,and a three-tier system architecture including data access,business logic and user interface was proposed.Functional model and data model of the system were constructed,and a rescheduling system which took plan management as core and process as support was developed.The system includes process data management,material management,personnel management,manufacturing execution management,quality management and planning control functions.After being applied to the steam turbine assembly site,the efficiency and stability of plan adjustment have been improved,and the labor cost has been significantly reduced.