Projects Due-date Forecast Methods Research While The Amount Of Shared Resources Obtained Under Uncertainty

In modern industry mould industry plays a very important role. As a typical make-to-order (MTO) pattern, the manufacturing of mould is full of randomness, which takes a lot of problems such as too long period of manufacture, serious tardiness. And mould project due-date as one of mould enterprise core competitive ability can't be managed effectively by existing production management method. So it is very urgent to find a method which suitable for mould manufacturing pattern.Consider the main random factors during the mould production process, combined with limited resources theory predict the future production effectively by simulate the production scheduling with stochastic theory which provide for mould production management a new way. In such demands environment, this paper studied the mould project due-date forecast method based on Markov decision process and combined with workshop load control theory under the frame of MRCPSP. The concrete content as follows:First, this paper analyzed the multi-project parallel, multi-project resource sharing and other mould production features. Extracted several major factors from the production process based on production history data, and constructed the probabilistic model of these random factors.Secondly, based on probabilistic model of these random factors and the workshop load control theory, delimit the project due-date prediction Markov decision process model, which including the definition of the five elements:decision epoch, system status, action set, state transition probability matrix, the reward function, and full described the method to evolve projects based on this model.Then, to deal with the curse of dimensionality which generated during the projects evolution, presents ATC priority rule sequencing and Markov chain coarse evolution to optimize the system state space in the original algorithm framework. It's proving the effectiveness of these two methods by a comparative analysis of three compression samples.Finally, select a large domestic mould manufacturing enterprises as research object, developed a mould project due-date forecast system combining the research results of the front, and compute a projects group case by using these methods, which results show that these methods have some practical. At the same time, introduce the enterprise application in detail by using enterprise real data and operations, and give a summary to the application effect.In this paper, we considered Markov decision process theory and relevant algorithms, construct a random environment for mould project due-date forecast model. Design, develop and implement a set with mould project due-date forecasting software, which put theory into practice and provide decision support for enterprise managers.