In order to fully exploit the relationship between temporal features in NOx emission data and improve the accuracy of NOx emission forecasting results, this paper proposes a NOx emission forecasting method based on a hybrid neural network model of convolutional neural network (CNN) and long short-term memory network (LSTM). Taking the historical data of a 300MW coal-fired boiler as a sample, the k-means clustering method is used to group NOx emission sample sets. Then the high-dimensional mapping relationship of NOx emission variables is extracted based on the convolutional layer and pooling layer of the CNN network to construct a high-dimensional time series feature vector, which is input the LSTM network. A NOx emission prediction model is established based on CNN-LSTM by training LSTM network parameters. Through the actual data verification of coal-fired boiler, the Mean relative percentage error of the proposed prediction model for training and testing samples are 1.76% and 3.85%, respectively, which are much lower than other models. The results show that the proposed NOx emission prediction model has significant advantages in terms of prediction accuracy and generalization ability.