Ultra-wideband (UWB) is an emerging technology with numerous applications, such as indoor positioning, object tracking, and wireless sensor networks. However, non-line-of-sight (NLOS) scenarios can cause significant performance degradation. In this paper, we propose a deep learning model specifically designed for accurately identifying various types NLOS UWB ranging signals for better error mitigation. We extract 13 feature parameters from low-cost UWB sensors to characterize channel quality. Our proposed deep learning model consists of a convolutional neural network module, a long short-term memory module, and a fully connected module for extracting spatial and time-domain features. Through field experiments, we demonstrate that our model achieves high accuracy and F1 score of 0.98 and 0.97, respectively, indicating its potential for accurately identifying different types of NLOS cases in UWB ranging.