CONVOLUTIONAL NEURAL NETWORK BASED INSPECTION OF BLADE-DEFECTS OF A WIND TURBINE

摘要

A computer-implemented method for determination of blade- defects is automatically carried out by a computing system (CS). In step S1), an image (01) of a wind turbine containing at least a part of one or more blades of the wind turbine is received by an interface (IF) of the computer system (CS). The image has a given original number of pixels in height and width. A step S2) basically consists of two consecutive steps S2a) and S2b) which are executed by a processing unit (PU )of the computer system (CS). In step S2a), the image (01) is analyzed to determine an outline of the blades in the image. In step S2b) a modified image (AI) is created from the analyzed image (01) containing image information of the blades only. Finally, step S3) consists of analyzing, by the processing unit (PU), the modified image (AI) to determine a blade defect (BD) and/or a blade defect type (BDT) of the blades. As a result, the blade defects (BD) and/or blade defect types (BDT) are output by the processing unit (PU).