Based on the IWOA-BiGRU-A-DKDE Ultra-Short-Term PV Power Interval Prediction

In order to quantify the uncertainty of PV power point prediction and provide more detailed prediction information for power system, an interval prediction algorithm based on IWOA-BiGRU-A-DKDE is proposed in this paper. First, to improve the prediction of Bidirectional Gated Recurrent Unit (BIGRU), an Improved Whale Optimization Algorithm (IWOA) is proposed by introducing the initialization of Improved Circle chaotic map, nonlinear convergence factor and adding inertia weight, and the IWOA-BIGRU model is constructed for point prediction. Then, based on the point prediction results of IWOA-BiGRU PV power prediction model, the adaptive bandwidth diffusion kernel density estimation (A-DKDE) is used to analyze the error statistics of point prediction. Finally, the photovoltaic power is predicted under three confidence conditions, and the corresponding power fluctuation interval is obtained. Experimental results show that the proposed method is more effective in fitting the error probability density and can obtain higher coverage and narrower interval width in interval prediction.