A Comparative Study on Diagnosing Wind Turbine Blade Fault Conditions using Vibration Data through META Classifiers

The primary purpose of this study is to characterize distinct blade faults using statistical parameters collected from quiver signals. The machine learning approach to classification includes feature extraction, feature selection, and feature categorization. The J48 decision tree approach utilizes to choose statistical characteristics from vibration signals extracted using the DAQ (Data Acquisition System). Finally, features are classified using meta classifiers that use random subspace classifiers and random committee classifiers. The accuracy and performance of other limitations are going to compare. A prototype model will develop that allows for accurate fault classification in a short period. This research work is unique because it employs meta-classifiers like Random Subspace and Random Committee Classifier to categorize wind turbine blades using vibration data quickly. With a computing time of 0.01 seconds, the Random Committee Classifier achieves an accuracy of 80%.