EFFECT OF SAMPLING FREQUENCY AND SAMPLE LENGTH ON FAULT DIAGNOSIS OF WIND TURBINE BLADE

Abstract

The purpose of this research is to determine the effect of sampling frequency and sample length on fault diagnosis of the wind turbine blade in order to optimise the cost as high sampling frequency leads to high cost. This study uses machine-learning approach followed by feature extraction and feature selection with J48 decision tree algorithm. In this take a look at, statistical features had been extracted from vibration signals with various sampling frequency and corresponding sample duration. With extracted features, function choice and category turned into achieved. With the help of J48 decision tree algorithm the optimal sampling size is obtained at sampling frequency of 4000Hz, which gives the classifier accuracy as 72.66%. The obtained results can be used with low cost accelerometer i. e. MEMS based accelerometer.