Monitoring and Diagnosis of Equipment Defects in Electric Power Substation Through Remote Infrared Temperature Measurement
Xiaoxue Ye, Rui Wen and Lin Jiang
The safe operation of the power grid can effectively be guaranteed by accurately detecting defects in electric power equipment in substations. This article briefly introduced the faults of substation equipment and infrared temperature measurement technology. In order to recognize faults and measure temperatures in infrared temperature images of substation equipment, the faster recurrent convolutional neural network (RCNN) algorithm was chosen. Simulation experiments were performed to compare the faster RCNN algorithm with back-propagation neural network and CNN algorithms in the laboratory. The three detection algorithms were tested for 20 days in a real substation. It was found that the faster RCNN algorithm was more accurate in recognizing and locating faults and measuring temperatures in substation equipment, both in simulation experiments and in the actual substation operation.
Keywords: Infrared temperature measurement, substation, defect detection, deep learning