Research On Model And Algorithm Of Distributed Capacity Allocation Problem In Random Environment

As an important decision making problem in the process of enterprise’s production and operation, the capacity allocation problem has a wide range of application background and significance. Recent years, because of economic globalization and transnational corporation of enterprise, in order to achieve continuous growth, rapid response to market demands and changes, more and more enterprises is adopting a distributed mode of operation, therefore, the capacity allocation problem based on distributed environment is imminent.On the other hand, due to the diversity and complexity of today’s society, the business decision makers are facing a growing number of uncertainties and asymmetric information in the decision-making process. Therefore, research on distributed capacity allocation problem in uncertainty environment with private information will be more practical significance, at the same time, the development of computer and communication technologies has also makes the exchange of information more convenient and frequent in the private information environment. Based on this, we study the distributed capacity allocation problem with random factors and private information.Based on the review on distributed decision-making theory and algorithms, stochastic programming theory, lagrangian relaxation method, we study two kinds of distributed capacity allocation problem about stochastic demand and processing time with private information. The main work is as follows:First, we give the distributed capacity allocation problem model with stochastic demand and processing time respectively by the theory of stochastic programming. Then we give the organization and facility two-tier structure model with private information based on the theory of lagrangian relaxation method and the Taylor expansion.Secondly, we update the Lagrangian relaxation multiplier by the sub-gradient optimization method. Then we give the solution algorithm, example and experimental analysis. Experimental results show the effectiveness of the algorithm. Based on the two cases of information completely unknown and some of the information are known, we get the results show that the known part of the information closer to the optimal solution under the centralized decision-making.Finally, we use improved deflected sub-gradient method to improve the previous methods and simulate by random data. Based on the comparative analysis about large data Improved deflection gradient method performance better than the general gradient method on the algorithm convergence.The consolidated results show that this study can as a strong theoretical basis for enterprise’s distributed capacity allocation in uncertain environment with private information.