Wind turbine gearbox fault prognosis using high-frequency SCADA data

Abstract Condition-based maintenance using routinely collected Supervisory Control and Data Acquisition (SCADA) data is a promising strategy to reduce downtime costs associated with wind farm operations maintenance. New approaches are continuously being developed improve the condition monitoring for turbines. Development of normal behaviour models popular approach in studies SCADA data. This paper first presents data-driven framework apply an artificial neural network turbine gearbox prognostics. A one-class support vector machine classifier, combining different error parameters, used analyse model develop robust threshold distinguish anomalous operation. detailed sensitivity study then conducted evaluate potential high-frequency The results based on operational from one show that, compared conventionally 10-min averaged data, use valuable as it leads improved prognostic predictions. High-frequency provides more insights into dynamics components can aid earlier detection faults.