Research On The Model And Algorithm Of Distributed Inventory Scheduling Based On Cloud Computing

Inventory Scheduling is an important component of logistics management. With the rapid development of e-commerce, the feature of inventory scheduling has changed from local, independent to regional and highly information sharing which reflected in the widely used of distributed inventory scheduling. The distributed inventory scheduling achieved a reasonable allocation of the regional goods inventory, significantly reduced inventory management costs through a warehouse information center. But the cross-regional sharing of logistical information and the efficiency and accuracy of the algorithm is always a big problem which need to be deal with. Cloud computing is a popular Big Data solutions in recent years. This paper uses Hadoop cloud computing platform and knowledge of parallel algorithm design to conduct research and experiments about the modeling of distributed inventory scheduling model, the algorithm and its performance optimization. The main contents are as follows:(1) Establish hierarchical control distributed inventory scheduling model based on cloud computing. For the defect of distributed inventory scheduling model under traditional centralized control and decentralized control, introduced the PaaS(Platform-as-a-Service) and deployed the inventory scheduling system on the cloud platform. Design a distributed inventory scheduling model under hierarchical control with distributed information sharing capabilities of the cloud platform. This model combines the advantages of traditional centralized control and decentralized control, full use of the value of shared inventory information.(2) Design a improved HS(Harmony Search) algorithm based on cloud computing to solve the model. For the defect of the traditional optimization algorithm of distributed inventory scheduling models performed poorly on the global search, select HS as the algorithm to solve the model, which has strong global search capability among the heuristic algorithm, and makes the algorithm parallelization to propose the dynamic parameters harmony search parallel algorithm. Establish harmony matrix library for stock subsystem of each region to make partial optimization. Then, make global optimization by the warehousing information center, achieve the inventory management in the regional inventory information-sharing environment. Experiments show that the algorithm for solving the distributed inventory scheduling model can escape from local optima faster to find the better global optimum, so it has practical value.(3) Load balancing optimization for Hadoop. For the efficiency of the algorithm executed did not meet expectations in the simulation, we research the task scheduling mechanism of MapReduce on Hadoop, propose DPLB(Dynamic Priority Load Balance) optimal scheduling algorithm. This algorithm use heartbeat message periodically sent between TaskTracker and JobTracker to design a dynamic priority scheduling feature quantity, solve the problem of imbalanced node load effectively during the execution of the task, improve the efficiency of solving the model.The research shows that the dynamic parameters harmony search algorithm designed for hierarchical control distributed inventory scheduling model based on cloud computing can get higher accuracy and convergence speed, and it can further improve the optimizing efficiency combined with the dynamic priority load balance platform optimization algorithm, so it has high significance of research and application.