61 China Communications ● Supplement No.2 2014 PV Power Short-Term Forecasting Model Based on the Data Gathered from Monitoring Network ZHONG Zhifeng 1 , TAN Jianjun 2* , ZHANG Tianjin 1 , ZHU Linlin 1 1 School of Computer and Information Engineering, Hubei University, Wuhan, 430062, Hubei Province, P. R. China 2 School of Information Engineering, Hubei Minzu University, Enshi, 45000, Hubei Province, P. R. China Abstract:The degree of accuracy in predicting the photovoltaic power generation plays an important role in appropriate allocations and economic operations of the power plants based on the generating capacity data gathered from the geographically separated photovoltaic plants through network. In this paper, a forecasting model is designed with an optimization algorithm which is developed with the combination of PSO (Particle Swarm Optimization) and BP (Back Propagation) neural network. The proposed model is further validated and the experiment results show that the predication model assures the prediction accuracy regardless the day type transitions and other relevant factors. In the proposed model, the prediction error rate is worth less than 20% in all different climatic conditions and most of the prediction error accuracy is less than 10% in sunny day, and whose precision satisfies the management requirements of the power grid companies, reflecting the significance of the proposed model in engineering applications. Keywords: grid-connected PV plant; short-term power generation prediction; particle swarm optimization; BP neural network I. INTRODUCTION Alongside the development of social productions, the demand for energy is also increasing which in turn in leads to both the worldwide energy crisis and prominent increase in the environmental pollution. Former incidents of the Soviet Chernobyl and Japan’s nuclear disasters caused by earthquake, highly questions the confidence level and dependability ratio of the nuclear power across the world. Many countries accordingly adopted significant policy shifts specifically to accelerate the development of the renewable energy. Solar energy is an important clean source of renewable energy. Over the past three decades, the utilization of the solar energy particularly for the photovoltaic technologies has gained rapid development. The random nature of the photovoltaic generation incurred by its fluctuant and intermittent power supply, impacts the power grids. Thus, power grid companies tend to strengthen their network information regarding the communication of the photovoltaic power plants in order to strengthen the supervision of each photovoltaic plant by forming a large-scale distributed PV power network, as illustrated in the following Fig.1. And the power grid companies are attempting to balance the generating capacities of various power plants, especially the generating capacities of the non- clean energy sources, by relatively predicting the photovoltaic energy based on the multiple information collected through the network. PV forecasting based on the network data can anticipate changes in the power generation in order to dispatch reasonable power generation capacity and to effectively utilize the resources. This improves both the security and stability of the photovoltaic power grids with the appropriate allocations of power generation capacities to the power grid companies. II. COMPARISON OF THE MULTIPLE TRADITIONAL PREDICTION ALGORITHMS Forecasting methods of the photovoltaic generation are evident in a large number. Directed prediction is SIGNAL PROCESSING FOR COMMUNICATIONS