Short-Term Forecasting and Uncertainty Analysis of Photovoltaic Power Based on the FCM-WOA-BILSTM Model

Aiming to solve the problem that photovoltaic power generation is always accompanied by uncertainty and short-term prediction accuracy of (PV) not high, this paper proposes a method for forecasting (PPF) analysis using fuzzy-c-means (FCM), whale optimization algorithm (WOA), bi-directional long memory (BILSTM), no-parametric kernel density estimation (NPKDE). First, principal component (PCA) used reduce dimensionality daily feature vector, then FCM divide weather into four categories: sunny, cloudy, rainy, extreme weather. Second, WOA train hyperparameters BILSTM, finally, optimized were construct WOA-BILSTM model types samples after clustering. The NPKDE was calculate probability distribution PV errors confidence intervals PPF. RMSEs FCM-WOA-BILSTM are 2.46%, 4.89%, 1.14% rainy types, respectively. simulation results calculation example show compared with BP, LSTM, GRU, PSO-BILSTM, FCM-PSO-BP models, proposed has higher under various which verifies effectiveness method. Moreover, can accurately describe forecast errors.