Fault Diagnosis of Wind Turbine Blades Through Vibration Signal Using Filtered Cultivation Data: A Comparative Study

Abstract

Due to the excellent wind resource and benefits of reducing land usage and visual impact concerns, wind turbines are installed more frequently in isolated onshore and offshore places. A wind turbine's rotor blades are vital in converting wind energy into electricity. The damage to the blades influences the power generation and turbine shutdown. In addition to the ongoing push to reduce the cost of wind energy, condition monitoring is currently generating a lot of attention since it is one of the most excellent solutions for maintenance problems. A pattern recognition system in machine learning approaches can detect and diagnose the faults in wind turbine blades. This proposed study demonstrates the effectiveness of machine learning models in identifying blade faults using filtered and unfiltered vibration signals. The logistic regression model using the resample filter-based vibration signal shows the best classification accuracy of 99.75 percent in 0.69 seconds.