Power plant power equipment overheating fault detection method based on visual recognition technology

Abstract

The application provides a power plant electric power equipment overheating fault detection method based on visual recognition technology, which can accurately detect the overheating fault of the power plant electric power equipment and improves the efficiency and accuracy of fault detection. The method comprises the following steps: S100, collecting image data of the power plant electric power equipment by using an infrared thermal imager; S200, pre-processing the collected image data, including denoising, enhancement and other operations; S300, segmenting the target region of the pre-processed overheating image by using an improved VLSBC level set contour model with a fusion bias field; S400, identifying the fault image region by comparing the support vector machine (SVM) algorithm; and S500, calculating the similarity value between the two by using the cosine similarity method, comparing the characteristic of the known fault mode, judging whether the equipment has a fault and the type and degree of the fault, and triggering the corresponding alarm mechanism. The application automatically learns and extracts the fault characteristics in the image by using the deep learning technology, thereby improving the efficiency and accuracy of feature extraction.

Images