Wind-turbine Blade Inspection Using Impact-Sounding Module and Acoustic Analysis

Wind is quickly becoming one of the most important renewable energy sources in the United States and around the world. One of the biggest challenges in wind industry is a constant need for inspection and maintenance of wind blades. This paper addresses the issue of inspecting subsurface delamination of blade using a mobile robot equipped with an acoustic impact-sounding system and RGB-D camera. Using a simple solenoid as a tapping device to collect acoustic data, we first use an Adaboost model to detect the delaminated area. Then, we use a Random Forest method model to perform region classification using the normal acoustic data. The acoustic analysis proves that frequency density provides a higher contrast between the delaminated region and the normal region than power spectral density method. Furthermore, We performed comparative experimental study on defects detection and area classification. Results show that our algorithm can achieve 88.57% accuracy in delamination recognition, and 78.33% in accuracy for region classification.