Optimization Solution Research For Decentralized Project Planning And Scheduling In OKP Enterprise

OKP (One-of-a-Kind Production) is a customer-driven modern productionmode, and most of the high level strategy productions, such as shipbuilding, voyageequipment manufacturing, belong to this mode. The OKP production has thecharacteristics of organized by projects, designed and manufactured according tothe orders, small batch of product, inter-organization coordination, andmulti-project parallel. All of these characteristics cause the project planning andscheduling in OKP enterprise has the complexity not just of traditional projectscheduling problem, and also the features and complexities such asinter-organization, decentralized, dynamic and coordinative decision making. Mostof the existing research works on the project planning and scheduling problem inOKP enterprise focus on the models and optimization solutions from the viewpointof centralized decision making; on the other hand, the achievements such as marketbased coordination mechanism in the research domains of game theory andinformation economics have much value for the project planning and schedulingproblems which has the feature of decentralized decision making in OKP enterprise.Consequently, it is necessary to study the decentralized characteristics of the projectplanning and scheduling in OKP enterprise, and research the optimization solutionfor decentralized project planning and scheduling, to coordinative operation ofmultiple organizations and projects, shorten project makespan and enhance theorder delivery ability.The main study work in this thesis contains the following parts:(1) Firstly, analyze the nature characteristics of decentralized project planningand scheduling problems in OKP enterprise. Compare the ideas between centralizedand decentralized solutions for project planning and scheduling problem, andpropose the features and modeling method. Then, based on the analysis above andtypical problems in OKP enterprise, propose the express method for decentralizedproject planning and scheduling problem. From the express method, we couldcompare and analyze different types of decentralized problems, models, solutionsand performances. Meanwhile, map and locate the work in this thesis and existingresearch based on the expression to guide the future research. (2) Study the solution for inter-organization order coordinative acceptance andcapacity planning. Focused on the three-component coordination order acceptanceand planning participated by the erection enterprise, customer and strategy supplier,analyze the influence by the dynamic candidate orders and potential order, andpropose local decision making model and three-component staged negotiationmechanism. Focused on the non-equivalence concession problem inthree-component structure, propose majority combined concession strategy andmulti-issue adjustment strategy for decentralized local decision makings. And thenpropose staged negotiation solution for three-component multi-issue decentralizedcoordination, and focused on the deficiency of stochastic programming for highuncertain problem, propose three different heuristic rules and present orderacceptance and capacity planning solution based on approximate dynamicprogramming. The computational results showed that the models and solutions hadbetter performance for order acceptance and planning problem than othernegotiation solutions.(3) Focused on the parallel multi-project scheduling problem which theseprojects have relationship of shared resource demand, analyze the influence by thedecentralized decision-making and coordination, and study the characteristic ofactivity duration elasticity for the project optimal makespan. Propose MAS basedmulti-project coordination model and project scheduling model with durationelasticity. Focused on the competition and coordination for shared resources amongmultiple projects, propose the shared resource combinatorial auction model andcombinatorial auction based decentralized multi-project scheduling andcoordination mechanism, and propose tabu search based project scheduling optimalalgorithm. The combination of project local optimization and multi-projectcoordination realize more reasonable shared resource allocation and optimalaverage project makespans. The computational results show that the decentralizedsolution has higher efficiency for multi-project scheduling than centralized method.(4) Study solutions for the robustness oriented decentralized project reactivescheduling problem with rescheduling added cost. Focused on the problem whichmultiple projects and multiple resource subjects participate in, introduce theresource exchange mechanism among multiple subjects to raise efficiency andrescheduling robustness. Propose MAS based decentralized project reactive scheduling model, and present combinatorial exchange based decentralized projectreactive scheduling and coordination mechanism. Introduce and expand theOR-XOR bidding language for the auction/exchange among multiple buyers andmultiple sellers to reduce trade cost of decision subjects. The computational resultsshow that the combinatorial exchange mechanism could raise scheduling robustnessand reduce added cost for project reactive scheduling.(5) Based on the strategies, models and algorithms in this thesis design anddevelop one coordinative project planning management system for shipbuildingenterprise. Faced the actual problems in Huanghai Shipbuilding Limited Companyand other master shipbuilding enterprises in China, apply and test the system andsolutions to examine the performance of the theories in this thesis.