As the core component of the Tunnel Boring Machine (TBM) rock-breaking function, disc -cutters directly affect the service life and construction efficiency of the TBM. Accurately predicting the wear status of disc -cutters is critical to efficient reel replacement decisions. However, due to the complex environment of cutters, both manual inspection and traditional sensor detection are subject to strong interference, which reduces efficiency and effectiveness. To solve this problem, a combined method LSTM-CNN based on long short-term memory (LSTM) network and convolutional neural network (CNN) is proposed, which predicts the wear status of the disc-cutters based on vibration dataset. The high-sensitivity vibration sensor is used to collect signals in the rock-breaking test, and then the deep learning model is used to eliminate interference signals to extract effective features, and the prediction of the three kinds of disccutters wear status (normal, uniform -wear failure and angled wear failure) is realized. Comparing the LSTM-CNN model with support vector machine (SVM) and traditional LSTM, the results show that the LSTM-CNN outperforms the other two models.