Wind Speed Prediction Model Based On Improved HCMAC With Credibility Assignment

Development and utilization of new energy is an important energy strategy of China and even the whole world in21th Century. Wind energy is a clean, renewable energy, has attracted more and more attention, and become one of the new energy. At present, the main form of exploitation and utilization of wind energy is a large-scale wind farms. Because the characteristics of the wind is intermittent, random inertia and uncertainty, wind farms connected to power grid bring new challenges to optimal operation of power systems, constraining the development of large-scale wind power. Accurate prediction of wind power generation power helps dispatchers on grid normal operation. Short-term wind speed forecast is one of the effective ways to solve the problem.By the support of the National Natural Science Foundation of China (No.51277127), the paper put forward Improved Hyperball Cerebellar Model Articulation Controller with Credibility Assignment (CA-IHCMAC). The algorithm is based on the storage unit reliability size, to adjust the storage unit according to the reliability allocation formula weight. Based on the wind speed prediction as the research object, using CA-IHCMAC neural network to build the wind speed prediction model for the future1h.The main contents of this paper are shown as follows:(1) The background and the significance of that research, analyzed the present situation of research on wind speed forecasting modeling, summarizes the main methods currently on the wind speed prediction modeling, and wind speed forecasting problems;(2) This paper introduces the concept of the Hyperball Cerebellar Model Articulation Controller, designed a kind of Hyperball Cerebellar Model Articulation Controller with Credibility Assignment. Analysis of the LSM weight tuning algorithm in weight adjustment process of storage unit, the adjustment of the storage units back will produce an impact called "corrosion phenomenon" on those already learned weights. In order to avoid the "corrosion phenomenon ", this paper presents a neural network weight tuning based on algorithm of reliability allocation. The algorithm is based on storing excitation unit live times the storage units credibility, to adjust the storage unit according to the storage unit reliability weights. According to the reliability formula, the more the number of memory cell activation, its reliability is higher and error adjustment is smaller.(3) In order to avoid the blindness of selecting reliability allocation hyperball neural network based on parameters, this paper joined the particle swarm optimization algorithm and researched on the optimization process of particle swarm optimization CA-HCMAC. The simulation results show that, with the parameters of particle swarm algorithm will be in short time and have higher prediction accuracy.(4) To determine the number of input variables of the neural network. In this paper, taking Shanxi Shenchi wind speed data as training samples to train the neural network, the data recorded the wind farm. This paper uses the time series model to analyze the inherent law of the future wind with its proximity historical wind speed and to find the internal relation between the wind data and the predicted wind speed. To determine the CA-HCMAC number of input variables, this paper used the stationary time series model established of time series.(5) A wind speed forecasting model. It is build on Hyperball Cerebellar Model Articulation Controller with Credibility Assignment. Through the MATLAB simulation platform, the simulation results show that, the hyperball cerebellar neural network based on the reliability allocation helps to increase the feasible and effective of the prediction of wind speed.