Abstract: In underground backfilling technology, long-distance pipeline transportation of gangue slurry is prone to problems such as sedimentation and blockage. At present, point-based acquisition and local observation methods are mostly used to monitor the pipeline transportation state, which makes it difficult to achieve continuous coverage and accurate localization of anomalies along long-distance pipelines. To address this problem, Distributed Acoustic Sensing (DAS) based on phase-sensitive optical time-domain reflectometry was adopted to achieve full-field continuous monitoring of blockage conditions in gangue slurry transportation pipelines. A gangue slurry transportation pipeline blockage experimental platform was established to simulate normal transportation as well as blockage conditions of 20%, 40%, and 60%. DAS optical fiber was used to collect vibration signals along the pipeline. A dual-branch identification model was constructed using One-Dimensional Convolutional Neural Network (1DCNN), Long Short-Term Memory (LSTM), and Cross Attention (CA) mechanism. The energy and power spectral density of DAS vibration signals were used as features to classify pipeline blockage conditions. The experimental results showed that the model achieved accuracy, recall, and F₁ score of 91.27%, 93.10%, and 92.28% for normal pipeline conditions, and 92.72%, 92.06%, and 92.39% for different blockage conditions, respectively. The performance was superior to that of DAS vibration signal identification models such as multi-scale convolution combined with a hidden Markov model. Based on the proposed method, a DAS monitoring and intelligent identification system for gangue slurry transportation pipeline blockage was developed, which realized visualization of vibration signal feature analysis and a B/S network service application for pipeline blockage identification.