A Study On The Task Scheduling Strategy Of Power Cloud Data Center

Building an integrated information platform is the development trend of power enterprise information construction and the foundation support platform which goes through all six parts of the smart grid. With the construction and development of integrated supervising platform of intelligent substation and communication and information platform, the electric power data shows an explosive growth. But the existing data centers in electric power corporation have a low efficiency when processing massive amount of data. Therefore, they can’t meet the electric power users’ needs for multi-quality of service. The core technologies of cloud computing are massive distributed data processing, virtualization and parallel programming models. With those technologies and its low cost, cloud computing can better meet the requirements of construction of the smart grid data center platform.After introducing the current situation of electricity cloud, data center and task schedule of cloud computing, this thesis invents a cloud platform for electric power system. As the customers of cloud service, cloud computing users’ satisfaction degree has a great influence on the development of cloud computing. This thesis analyzes factors influencing the users’ satisfaction and proposes to measure the satisfaction using fuzzy subjection function. Based on the function, with the optimization objectives being to maximize customer satisfaction degree and minimize cloud computing energy consumption, a cloud-computing scheduling optimization model based on customer satisfaction degree is established, in which the basis of customer value evaluation is decided by the weight coefficient of satisfactory function in the model.As cloud-computing scheduling optimization model based on customer satisfaction degree is a multi-target optimization model, the thesis adopts linear weighting method to convert multi-target functions in the model into single-objective functions, and applies improved genetic algorithm to solve the model. Finally the thesis uses examples to verify the effectiveness of the model and the algorithm.