Research And Application Of Scheduling Strategy In Power Cloud

With the rapid development of smart grids,state monitoring technologies for power systems are moving toward high sampling rates,large-scale and large-capacity storage.The traditional power monitoring system has shortcomings in the high concurrency of information integration,storage and acquisition equipment.The original parallel Hadoop scheduling algorithm can not meet the real-time requirements of power collection equipment.Therefore,it is necessary to design a more reasonable power cloud scheduling.The algorithm meets the real-time requirements of power dispatching.This article focuses on how to properly allocate and control computing and storage resources and ensure that jobs are executed in the correct order.The main contents are as follows:Firstly,the research status of cloud computing and Hadoop platform at home and abroad is analyzed.The HDFS distributed file system and MapReduce programming model of Hadoop are deeply studied.The concept of power cloud is introduced,and the remaining computing and storage resources in the smart grid are integrated into the resource pool.Based on the form,the feasibility of FIFO,Capacity and Fair algorithms applied to power acquisition equipment for parallel data collection is studied.The research focuses on the high concurrency of data transmission in acquisition equipment.Then,for the parallel monitoring of power equipment monitoring,the standard firefly algorithm was improved,the calculation formula between individuals was redefined,the position update formula was improved,and the group theory was combined to improve the diversity of the population.In the MATLAB environment,the standard firefly algorithm and the comparison experiment based on the multi-group firefly improvement mechanism algorithm were used to verify the effectiveness of the MFA_IEM algorithm.Finally,MapReduce is used to parallelize the firefly algorithm,and the service selection method based on MFA_IEM algorithm is proposed.The coding method of firefly in MapReduce mode is given,and the parallel computing of small-scale multi-group is realized.The specific Map function and Reduce are given.function.Through the original scheduling algorithm and improved scheduling algorithm,the testand scheduling of WordCount on the built power cloud platform are compared in terms of CPU utilization,job response time,scheduling accuracy,etc.,and the performance of the MFA_IEM algorithm is verified.Some scheduling algorithms and show good scalability.