Continuous RBM Based Deep Neural Network for Wind Speed Forecasting in Hong Kong

摘要

The wind speed forecasting in Hong Kong is more difficult than in other places in the same latitude for two reasons: the great affect from the urbanization of Hong Kong in the long term, and the very high wind speeds brought by the tropical cyclones. Therefore, prediction model with higher learning ability is in need for the wind speed forecast in Hong Kong. In this paper, we try to employ the Deep Neural Network (DNN) to solve the time series problem of wind speed forecasting in Hong Kong since it is believed that Neural Network (NN) with deep architectures can provide higher learning ability than shallow NN model. Especially, in our paper, we use the continuous Restricted Boltzmann Machine (CRBM) to build the network architecture of the DNN. The CRBM is the continuous valued version of the classical binary valued Restricted Boltzmann Machine (RBM). Compared with the Stacked Auto-Encoder (SAE) model applied in our previous study, this CRBM model is more generative, and therefore more suitable for simulating the data in wind speed domain. In our research, we employ the DNN to process the massive wind speed data involving millions of hourly records provided by The Hong Kong Observatory (HKO). The results show that the applied approach is able to provide a better features space for computational models in wind speed data domain, and this approach is also a new potential tool for the feature fusion of continuous valued time series problems.