| The centralized control center system of the wind farm integrates real-time monitoring,early warning,wind power statistical analysis,power forecasting,insurance,WEB release and other functions,realizing the new management and operation mode with fewer field-stationed engineers and fewer or even no people on duty.However,as the scale of the wind farm expanse,it produces a huge amount of data every day with more collections of points and higher frequency,which causes problems such as a decrease in the data system performance and the slow to query and calculate so large amount of data after years of accumulation.The centralized control center established on a cloud platform in the security zone III satisfies the remote control needs of the wind farm in the future by realizing the monitoring functions and maximizes data sharing and resource integration by enhancing data storage and analysis capabilities.It lays the foundation of the future migration from cloud platform technology to the security zone I by providing technical support of remotely operating the wind farm and offering a basic platform for future intelligent application expansion.Therefore,Based on the centralized control center established on a cloud platform in the security zone III,this paper will mainly study from the following aspects.(1)The overall structure diagram,security partition,centralized control center network,remote communication network,wind farm side network and wind power data acquisition method have been designed while the subsystems of side monitoring,wind power prediction,energy management,fault expert diagnosis and so on have been completed for the wind farm control center.(2)The problems such as a decrease in the data system performance and the slow to query and calculate so large amount of data after years of accumulation.Distributed file system HDFS,distributed computing model Map Reduce and database HBase and Yet Another Resource Negotiator(YARN)and other related technologies have been studied in depth.Based on the cloud platform designed by the centralized control center,which can achieve a rapid data storage and remote wind and power monitoring: boost station monitoring function,wind monitoring function,production information,photon alarm,AGC/AVC monitoring function,reporting function,alarm function and data mining function.(3)Hadoop cloud platform has been built in the laboratory and the comparison of data import by the cloud platform with that of the traditional database has been performed to verify the rapidity of the cloud platform storage.Use locally weighted linear regression to model wind power curve with the distributed computing model Map Reduce to design a polynomial function in parallel and predict the actual value for a wind power in half a month.The accuracy of the results reaches 96.0128% and proves the validity of this model in the evaluation of the wind performance.And compared with the polynomial algorithm and the classical algorithm SVM and proves that the local weighted linear regression is more suitable for the modeling of the actual wind power curve by the comprehensive comparison.After Compared with the fitting of a polynomial the classical algorithm SVM,it proves that locally weighted linear regression is more suitable for modeling the actual wind power curve.Finally,through a lot of experiments,the parallel performance indicators(efficiency,speedup,expandability)of the cloud platform has been tested and analyzed,which further verified the advantages of the cloud platform. |
