Abstract: Thermal power generation is still the major power generation mode in China. Since the quality of coal quality directly determines the safe production and economic benefits of power plants, net calorific value as received basis has become one of the key indicators of coal quality. Regarding the current problems such as complicated measurement procedures of coal calorific value and limitations of the general method in situations where real-time monitoring is required, this paper proposes a convenient and fast calorific value estimating algorithm based on hyperspectral image and convolutional neural network. Firstly, the image data of coal is collected by the hyperspectral data acquisition system. After Gaussian low-pass filtering and principal component analysis, the noise and data redundancy between spectral channels are eliminated. Then, the smooth training and test data are obtained by using the neighborhood average filtering and a 7-layer convolutional neural network is established. Finally, the effectiveness of the proposed method is verified by experiments. The results show that the method has high prediction accuracy.