Abstract: Interval prediction method can reflect the possible range of photovoltaic power， and provide more abundant forecasting information than point prediction method. An interval prediction model based on Radial Basis Function （RBF） neural network is proposed to directly output the prediction interval of photovoltaic power. In order to optimize the performance of the output interval of the model and avoid the problem of choosing penalty coefficient， an improved prediction interval optimization model considering the bias information of the interval prediction is constructed and solved by particle swarm optimization （PSO） algorithm to obtain the optimal RBF neural network output weights， and improve the reliability and accuracy of the prediction interval. By comparing the prediction results of the traditional interval optimization model and the improved interval optimization model， it is found that the improved interval optimization model can obtain more accurate prediction interval of photovoltaic power， and can provide more accurate auxiliary information for dispatching decision-making.

Key words: photovoltaic power, interval prediction, radial basis function , （RBF） neural network, particle swarm optimization（PSO）