A model-based approach to wind turbine condition monitoring using SCADA data

Modern wind turbines are complex aerodynamic, mechanical and electrical machines incorporating sophisticated control systems. Their design continues to increase in size and they are increasingly being positioned offshore where the environment is hostile and where there are limited windows of opportunity for repair and maintenance activities. Condition monitoring is essential offshore if Wind Turbines (WTs) are to achieve the high reliability necessary for sustained operation. Contemporary WT monitoring systems already provide vast amounts of data, the essential basis of condition monitoring, much of which is ignored until a fault or breakdown occurs. This paper presents a model-based approach to condition monitoring of WT bearings. The backbone of the approach is the use of a least squares algorithm for estimating the parameters of a discrete time transfer function (TF) model relating WT generator temperature to bearing temperature. The model is first fitted to data where it is known no problems exist. It is then used in predictive mode and the estimates of the bearing temperature are compared with the real measurements. The authors propose that significant discrepancies between the two are indicative of a developing problem with the bearings. The promising experimental results achieved so far indicate that the approach is viable.