Abstract: The application of major deep learning models in time series data processing and local feature extraction is introduced.And some strategies for ensemble learning are also put forward so as to overcome the limitations of single models.In addition, the latest progress of deep learning in the fields of signal search and weather forecast is discussed.On the other hand, an overview of the forecast capacity of seasonal to sub-seasonal dynamic models as well as the technological routes employed in operational extended-range weather forecast in China is presented.And the potential application of deep learning technology in extendedrange forecast is discussed in a bid to offer some enlightenment for the development of extended-range forecast in China.