Sensor selection based on principal component analysis for fault detection in wind turbines
Document typeConference report
Rights accessOpen Access
Growing interest for improving the reliability of safety-critical structures, such as wind turbines, has led to the advancement of structural health monitoring (SHM). Existing techniques for fault detection can be broadly classified into two major categories: model-based methods and signal processing-based methods. This work focuses in the signal-processing-based fault detection by using principal component analysis (PCA) as a way to condense and extract information from the collected signals. In particular, the goal of this work is to select a reduced number of sensors to be used. From a practical point of view, a reduced number of sensors installed in the structure leads to a reduced cost of installation and maintenance. Besides, from a computational point of view, less sensors implies lower computing time, thus the detection time is shortened. The overall strategy is to firstly create a PCA model measuring a healthy wind turbine. Secondly, with the model, and for each fault scenario and each possible subset of sensors, it measures the Euclidean distance between the arithmetic mean of the projections into the PCA model that come from the healthy wind turbine and the mean of the projections that come from the faulty one. Finally, it finds the subset of sensors that separate the most the data coming from the healthy wind turbine and the data coming from the faulty one. Numerical simulations using a sophisticated wind turbine model (a modern 5MW turbine implemented in the FAST software) show the performance of the proposed method under actuators (pitch and torque) and sensors (pitch angle measurement) faults of different type: fixed value, gain factor, offset and changed dynamics.
CitationPozo, F., Vidal, Y. Sensor selection based on principal component analysis for fault detection in wind turbines. A: European Conference on Structural Control. "EACS 2016 Proceedings". Sheffield: 2016.