Research On Wind Power Forecasting Algorithm Based On Feature Weighted Fuzzy Clustering Analysis

With the rapid development in recent years, wind power has accessed power system in large scale. However, the volatility, intermittent and instability of wind power bring serious challenges to the safety and stability of power system. Accurately forecasting wind power is one of the key technologies to solve the problems resulted from wind power accessing power system in large scale, which can help power system formulate corresponding scheduling control strategy in advance and ensure the safe and stable operation of power system. Therefore, the research on wind power prediction method is of great importance.Based on research object of a domestic wind power plant, wind speed, the distribution of wind direction, characteristics of wind power plant and wind electrical power are analyzes in this paper. On this basis, we further study wind power prediction method with higher precision. At the same time, the paper introduces cloud computing technology to power system, explores the smart grid load forecasting and discusses new development direction of optimal allocation of resources. The main contents of this paper are as follows.Firstly, a wind farm is taken as an object, and we research wind speed, wind direction, spatial distribution of wind power, wind power and wind electrical power characteristics. The results are shown that the probability distribution of wind speed is presented weibull distribution, the wind direction is shown as certain season features, wind fan is presented correlation of delay and spatial distribution, and wind power characteristics own the stochastic volatility and output characteristics with strong regularity.Secondly, the paper proposes an improved weighted fuzzy clustering algorithm and wind power short-term forecast method with the combination of Elman neural network model. Because the wind physical properties have different importance on identifying types of wind, we introduce weighting factor in the traditional FCM fuzzy clustering algorithm to comprehensively cluster wind type data sampling of history day. Then, the paper establishes dynamic Elman neural network model to cluster the results, which is used to predict wind power output value of the same clustering results in the target day. Using a domestic wind field to take simulation experiment, the measured data proves the superiority and practicability of the proposed method.Finally, the paper puts forward wind power prediction resource scheduling platform architecture based on cloud computing. It deeply expounds the key technology of cloud computing, wind power prediction resource management and resource scheduling mechanism based on cloud computing. Using cloud computing technology with Hadoop, we simulate resource scheduling of wind power prediction and computing services with virtual cloud computing, which verifies excellent cloud computing based resources optimization scheduling mechanism. Hence, we build cloud platform architecture to provide broad thinking and strong technical support for smart grid resource scheduling and load forecasting calculation.