Wind generating set blade fault judgment method and device

The invention belongs to the field of wind generating set blade fault judgment methods, and particularly relates to a wind generating set blade fault judgment method and device. The method comprises the following steps: firstly, automatically extracting Haralick texture features and SIFT key point features of a leaf surface; edge detection is carried out by applying a Frei-Chen operator, and edge information is converted into a weight, so that effective fusion of features is realized. A weighted feature fusion calculation model is further designed, and a weighted feature fusion database is constructed in combination with deep convolution features extracted by the SSD network. And finally, through comparing leaf image data collected in real time with a database, whether a fault exists in the leaf and the type and the position of the fault are rapidly and accurately judged. According to the method, the feature extraction capacity is improved through deep learning, automatic fault detection is achieved, manual i