Research On Scheduling Control And Optimization Method Of Marine Engineering Equipment Project

With gradual promotion for the China’s marine strategic status,marine engineering equipment(MEE)has become one key competence constrain to the marine resource exploitation.Meanwhile,the globalmarket demand for MEE continues to decrease,which leads to great pressure on the survival of MEE enterprises.Therefore,improving the capacity of production is the only way for MEE enterprises to deal with the new situation.Though the project scheduling management is one of the important factors that affect the production capacity of the MEE,there is no standard system in China,which leads to the phenomenon of extensive planning,poor flexibility in scheduling and unbalance of resource use;Additionally,the application of intelligent methods in the scheduling management process of MEE project(MEEP)is still blank,so there is a lack of efficient way in scheduling optimization and decision making.Therefore,to shorten the project duration,reduce the cost of the project,balance the enterprise resources,enhance the flexibility of the scheduling and improve the lean level,this dissertation studies the scheduling management mode of the MEEP and builds a multi-agent based scheduling management system model,then designs multi-objective optimization method and high performance calculation method oriented to for intelligent control of MEEP.On this basis,scheduling management system of MEEP is developed and verified.Regarding to the problem that scheduling management system in China’s marine industry is not perfect,after investigating and analysis on the status of the scheduling management of typical marine enterprises in China,the scheduling management process and scheduling control method of typical MEEP are obtained.Combined the idea of integration of design,construction,management and demand of intelligent control for project schedule,project scheduling management mode that satisfies the current situation and future development requirements of the marine industry enterprises is established.Concerning the problem of the blank of intelligent application in scheduling control of the MEEP,a scheduling management system model of MEEP using Multi-agent technology is established.The 2-stage scheduling control mechanism of MEEP based on Multi-agent is proposed,then a mechanism consisting of communication,collaboration,competition,and negotiation between agents is designed and the negotiation algorithms for two typical conflict problems are also presented,which above works lay the foundation for the realization of the intelligent control of the whole business process of the scheduling management of MEEP.In view of the intelligent planning and optimization of the push plan for the scheduling control of MEEP,the multi-objective optimization theory is applied.The multi-objective optimization model of time-cost-resource under resource constraint and multi-mode is established,then improved hybrid NSGA-Ⅱis designed based on NSGA-Ⅱ and Paretodominated SA.Combined with the characteristics of MEEP planning,heuristic population initialization strategy,acceptance algorithm of adaptive new execution mode and reset mutation method are proposed respectively.At last,through case simulation and comparative analysis,it is verified that the algorithms designed in this dissertation has better search ability and robustness in solving the push plan problem of MEEP.Aiming at the intelligent planning and optimization of the pull plan for the scheduling control of MEEP,the scheduling model based on Just in Time(JIT)is established.According to the characteristics of large-scale problem-solving in the process of scheduling control of MEEP,a department-based job grouping strategy and a stage-based job grouping strategy are proposed,which achieves the simplification of problem scale.Applying the parallel computing technology,the heterogeneous parallel computing system based on CPU/GPU and the corresponding algorithms are designed: constructing sub-problem-solving control layer,designing sub-problem negotiation algorithm based on Multi-agent;constructing the population behavior control layer,designing a cooperative evolution algorithm based on the confidence space;constructing the individual behavior control layer,realizing heterogeneous computing based on CPU/GPU.Finally,through case simulation,it is verified that the parallel computing system designed in this dissertation has the acceleration effect of more than 200 times compared with the conventional computing system,additionally,this method has the ability of real-time scheduling in solving large-scale pull scheduling problem of MEEP.To provide the application carrier and verification environment for the above research results,the service oriented architecture is adopted to develop the schedule management system of MEEP.The prototype system is demonstrated and operated in an enterprise of CSSC,which achieves excellent application effect.