Real-time running state monitoring of power equipment based on image recognition

It is a new method that monitoring the power equipments state based on image analyzing, and it can improve the automatic level of power system. Not only the video frequency would be monitored and recoded, but also the images analyzing and recognizing can be done, and it leads the incipient fault to be discovered on-line. The computer captures remote equipments images, calculates images' moment invariants as characteristic vector of recognition, recognizes power equipments by support vector machine (SVM). The system further analyzes images and find if images has convex hull, intensive stochastic noise, or false edges, sequently make a conclusion whether the running state of equipments is in order. Replace of scouting by people themselves, the use of image recognition in power system can timely find troubles and potential troubles of power equipments. So the power equipments can be inspected and maintains its working state, which would improve the running reliability of the unattended substation. (6 pages)