Research Of Task Scheduling System In Electric Power Cloud Computing

With the smart grid technology developing, power information system is facingso many issues, such as data massive, distributed, heterogeneous, complex, hard using,difficult maintenance, etc. However Cloud Computing is an effective solution to copewith the above problem by its’ powerful processing capabilities, dynamic, flexible,virtualization and service oriented properties. So we can build the Electric PowerCloud Computing-the private Cloud Computing of electric power system. InElectric Power Cloud Computing, task scheduling system is one of the most importparts, which can improve the concurrency of system, ensure a rational allocation ofusers’ tasks and resources, and improve the performance of power cloud. Thereforethis papers research the task scheduling system of Electric Power Cloud Computing,and major concerns two aspects: scheduling system model and scheduling algorithm.Basing on the research of basic theory and electric power information platform,this paper proposed a task scheduling system of Electric Power Cloud Computingbased on MapReduce. This system retains the advantage of MapReduce, andimproves the parallelism with tasks split and tasks preprocessing. And based on thescheduling system model, this paper proposed a task scheduling algorithm of ElectricPower Cloud Computing, and this algorithm used the Genetic Algorithm as theprototype to meet needs of different users with multidimensional constraints. At lastthis paper compare this algorithm with three different algorithm, and the comparingsimulation experiments result show that this algorithm is a very efficient taskscheduling algorithm.