A Fire Warning Technology for Power Equipment Based on SoC FPGA Stream Computing

Abstract

Before a fire, power equipment will have thermal defects, which can be shown by infrared thermography. The thermal image can be processed at the edge node without network transmission. Therefore, a fire warning technology for power equipment based on SoC FPGA stream computing is proposed. The technology applies the parallel acceleration capabilities of FPGA to improve terminal computing power. A priority queue is designed on the software side of the SoC FPGA to cache images of different priority. Parallelized gray-scale processing and abnormal temperature determination of image pixels are designed on the hardware side. Compared with software processing, this method has higher efficiency for the processing of streaming image data on terminal.