Research on Key Technology of Intelligent Detection of Elevator Cabinet Circuit Faults Based on Deep Learning

Aiming at the circuit faults in elevator control cabinets in complex design structures and variable operating environments, traditional fault detection methods usually rely on manual inspection and empirical judgments, which are inefficient and susceptible to the influence of human factors, resulting in some small faults that are difficult to be identified in a timely manner. To solve the above problems, this paper proposes an intelligent fault detection system based on deep learning. A convolutional neural network is used in combination with the YOLOv10 model, aiming to improve the automation and accuracy of fault detection. A customized dataset containing 10,000 fault images of elevator control cabinets was first constructed, covering five common fault types such as short wires, missing wires, missing plugs, exposed Y-connectors and exposed I-connectors. The creation of the dataset provides rich samples for model training and validation. In the experiments, the deep learning model based on PyTorch framework is used for training, the SGD optimizer is used for parameter updating, the initial learning rate is set to 0.01, and the cosine annealing algorithm is used for learning rate decay. The training results show that the model achieves an average precision mAP of 91.1%, a precision rate of 93.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>, and a recall rate of 90.1% on the test set. Especially in the detection of small targets, the system demonstrates strong robustness and accuracy, which significantly improves the ability to recognize circuit faults.