Addressing the limitation of point prediction in conveying adequate uncertainty information, this paper introduces a precise method for short-term photovoltaic power interval prediction. Initially, Pearson and Spearman double correlation coefficients are employed to assess the correlation between various meteorological factors and photovoltaic power generation. Subsequently, the Fuzzy C-means (FCM) clustering technique is utilized to categorize historical datasets into three distinct categories. Following this, we propose a QR-BiGRU hybrid model that integrates bidirectional gated recurrent units (BiGRU) with quantile regression (QR) model. Finally, we compare our model against QR-LSTM, QR-BiLSTM, and QR-GRU models using multiple evaluation metrics to assess their predictive performance. The experimental results indicate that the interval prediction approach presented in this study demonstrates high accuracy and offers valuable insights for scheduling within power departments.