Research Of Cluster Scheduling Algorithm In Cloud Computing Based On Logistics Data

As information technology develops quickly in past years, the technology of IOT(Internet of things) gets more and more attention. The combination of and logistics technology forms intelligent logistics industry. With the technology of RFID, GPS and so on, intelligent logistics achieves identifying goods automatically. The method of mining information from logistics data remains to be studied.This paper focuses on logistics data processing. Several classical data mining methods and machine learning algorithms are introduced into the application of logistics data processing, while some improvements are made to the algorithm for the sake of improving the capabilities of logistics data analysis and processing.Our major contributions are described as follows:(1) This paper focuses on K-means clustering algorithm for logistics data, understands of some problems existing in the K means clustering, such as classic algorithm selecting initial cluster centers randomly. To solve the randomness problem, the Prim minimum spanning tree is introduced to select the initial center of K-means clustering algorithm. This paper proposes an improved K-means clustering algorithm based on prim. The prim method is used to find the minimum spanning tree for the randomly generated points and can reduce the number of iteration and improve the clustering accuracy rate effectively without changing the simplicity of K means clustering algorithm.(2) This paper focuses on support vector machine algorithm for logistics data. To deal with the problem of machine learning of large data, stochastic gradient descent algorithm is applied to reduce training time and improve the training speed. This paper proposes an improved support vector machine based on stochastic gradient descent. The results of Simulation show that it can improve the training speed effectively without affecting the correct rate in a large number of data.(3) This paper focuses on cloud task-scheduling for logistics data. This paper proposes a green cloud task-scheduling algorithm(GCTA) based on the improved binary particle swarm optimization(BPSO). The major contribution of our work is avoiding matrix operations by using pipelined number for virtual machines and redefining position and velocity of particle. Simulation shows that the GCTA has less execution time, and reduces resource consumption accordingly.