Research On Wind Power Prediction System Based On Cloud Platform

During the period of "much starker choices-and graver consequences-in",power clean energy become the main propulsion task of our country electric power industry,China's wind power energy is gradually from major alternative electric energy into electric energy,as of 2016,China's total installed wind capacity has reached 168.73 GW,ranked first in the world.However,due to the natural wind speed change trend of great randomness and unable to manipulate,wind power base in our country the construction scale and production capacity in the energy industry as a whole body under the condition of increasing proportion,wind power from the active and reactive power balance on the scheduling of all caused great trouble to the power grid,the grid capacity of credibility is relatively low.To eliminate wind accounted for improve the power grid,the threat of power quality and power system stability in recent years,the wind power prediction technology got more research in our country,also more and more wind power prediction system was established gradually.Traditional wind power prediction system is mainly composed of the database server,power prediction,application servers,workstations,data interface and wind towers of five parts,its main principle is to use physical model projections on time output for wind turbines in the future,it needs to establish wind tower at the scene of the wind power production device,and the wind power field physical parameters for precise modeling,again by wind power conversion formula to predict the next moment,fan because of physical parameters of the model is complex,high precision requirements,process trival manifolds,for all kinds of time scale of the wind power of accurate prediction is very difficult,in addition,due to system maintenance manpower cost is high,the mesoscale meteorological data to a third party and wind towers rely on serious issues such as hardware construction,the working efficiency of the wind power prediction system is also difficult to meet the ideal requirements.For this,put forward a kind of wind power prediction system based on cloud platform,with wind power implementation of the statistical combination forecast model to predict the goal,in order to solve the traditional wind power prediction accuracy prediction system is not accurate,long construction period,difficult maintenance,high cost,the problem of shortage of work efficiency.Wind power prediction system based on cloud platform of multiple modeling methods to obtain the output prediction results,using intelligent weighting ensemble algorithms for processing,get higher prediction precision of the result data,at the same time meet the real-time model updating and the scene modeling approach to the needs of field situation.The system is composed of hardware system,software system and network data communication link.On the hardware,the system can be divided into three parts,such as the forecast cloud platform and the information display of the wind field side,and the network security equipment.Software,according to the functional design for data collection and storage module,the wind power prediction module,the downscaling forecast module,the data synchronization system and for the Web service information management system and so on several parts;The data communication link is designed according to the safety standard of the secondary side of the power system in accordance with the electrical supervision.Design,the system by introducing a cloud computing platform architecture based on Hadoop,for large scale numerical weather prediction downscaling operation and prediction model of training design parallel computing framework,in order to give full play to the cloud platform more powerful CPU computing power,enhances the working efficiency of the system;The real-time database of distributed deployment in the cloud provides reliable and high performance support for the storage and access of real-time mass data of wind farms.