Existing methods for artificial intelligence-based cable fault diagnosis lack the ability to handle the dynamic and differentiated development of faults

and they cannot adapt to new fault scenarios or enhance diagnostic capabilities over time. Therefore

this paper proposes a scalable diagnosis method for cable fault based on the deep convolutional neural network(DCNN)

incorporating the idea of knowledge distillation and attention mechanism(DCNN with knowledge distillation and attention

DCNN-KD-A). The knowledge distillation is introduced to retain the diagnostic information of the original faults by the DCNN model; meanwhile the classification loss of the newly-added faults is used to expand and modify the model

enabling it to have the capability for extensible fault diagnosis. To avoid the catastrophic forgetting problem in model training

a suitable fault classifier has been reconstructed. To address the issues of information loss and changes in knowledge distillation

the attention mechanism is utilized to enhance the model's feature extraction capabilities

thereby improving its scalability in diagnostic accuracy. The experiments are conducted using actual measured data and simulation data

and the experimental results verify the effectiveness and feasibility of the method proposed in this paper.