Identification of Incipient Faults in 10 kV Power Cables Based on SDAE-APO-LightGBM

The incipient faults in 10 kV power cables are of the type known as intermittent arc faults, and are characterised by a short duration and random occurrence. The identification of these faults represents a significant challenge. In order to address this challenge, an incipient fault identification method based on stacked denoising autoencoder-artificial protozoa optimizer-light gradient boosting machine (SDAE-APO-LightGBM) is proposed. Firstly, a stacked denoising autoencoder (SDAE) module is employed to perform dimensionality reduction and feature extraction on the phase current data containing incipient faults, thereby constructing a feature dataset. This dataset is then fed into a light gradient boosting machine (LightGBM) classifier, optimized through the artificial protozoa optimizer (APO) algorithm, resulting in the precise identification of incipient faults in 10 kV power cables. The experimental results indicate that the proposed method is highly effective in identifying incipient faults in 10 kV power cables, demonstrating both high identification accuracy and robust performance, and thus possesses considerable potential for engineering applications.