Deep learning-based water-light-wind power storage short-term scheduling method and system

The invention provides a water-light-wind power storage short-term scheduling method and system based on deep learning. The method comprises the following steps: calculating predicted output and output volatility of wind, light and electricity after a preset time interval; comparing the output fluctuation ratio of the wind, light and electricity with a set threshold value, if the output fluctuation ratio of the wind, light and electricity is greater than the set threshold value, entering the next step, otherwise, returning to the previous step after a preset time interval; collecting operation data of water-light-wind power storage according to a preset sampling period, and collecting data in a preset time length in total; combining the collected operation data into a matrix and carrying out normalization processing to obtain a normalized matrix; and transmitting the normalized matrix into a trained deep learning model to obtain a corresponding scheduling strategy. According to the method, by utilizing the de