Predictive maintenance of electrical equipment based on ResNet-101 and Faster R-CNN

Abstract With the advancement of industrial intelligence, the stable operation of electrical equipment becomes a core element in ensuring enterprise productivity. Predictive maintenance of electrical equipment is emerging as a key strategy for enterprise resource allocation. However, current predictive maintenance methods face challenges such as technical complexity and poor data quality, which lead to inaccurate prediction results. This paper raises a model that integrates Residual Network-101 and Faster Region-based Convolutional Neural Network. The model fully utilizes the ability of deep feature extraction and accurate anomaly detection to calculate precise results. Test results show that the model achieves a training accuracy of 99.3% and a validation accuracy of 99.6%, which are significantly higher than those of comparison models. In the classification and recognition test of electrical equipment, the recognition rate of device No. 2 reaches 99.12%, also notably outperforming other models. In conclusion, the proposed model effectively addresses the accuracy issues in current predictive maintenance methods and provides a new approach for predictive maintenance. This improvement enhances production safety and helps ensure efficiency in industrial systems.