In this paper, infrared sensor technology is applied to the monitoring of electrical equipment, and a mathematical model of infrared image processing is built. In addition, the infrared image recognition model of electrical equipment is combined with the improved CenterNet, and ResNeXt50, which has stronger feature extraction capability, is used as the backbone network of CenterNet to improve detection accuracy. An infrared dataset of electrical equipment is used to verify the effectiveness of the improved CenterNet model and the other two models. Finally, the faults in substation equipment are analyzed and studied for real cases applied in practical work as an example. The results show that the algorithm in this paper has an mAP value of 93.7% and a frame rate of 40/S, which can balance the detection accuracy and detection speed in the electrical equipment infrared scene. Compared to traditional electrical equipment detection methods, the detection method of infrared sensor imaging is more reliable, with a 90% accuracy rate, and the highest has reached 100%. The article analyzes the characteristics and relationship between the local heating and temperature increase of the fault point of the equipment concerned, summarizes the fault judgment method, and verifies the effectiveness of infrared sensor imaging technology.