This paper proposes an intelligent diagnosis method based on multi-source information fusion and one-dimensional convolutional neural network (MS-1DCNN) to address the limitations of traditional low-voltage power supply cable fault diagnosis methods, such as reliance on a single signal source, insufficient feature extraction capabilities, and weak anti-interference performance. By acquiring cable fault data through the signal injection method and constructing an MS-1DCNN model, deep features from different signal modalities are extracted using multi-channel parallel one-dimensional convolutional neural networks (1D-CNN). Feature fusion is achieved through a channel attention mechanism to enhance fault classification capabilities under complex operating conditions.