It is common for wind turbines to be installed in remote locations on land or offshore, leading to difficulties in routine inspection and maintenance. In addition, wind turbines in these locations are often subject to harsh operating conditions. These challenges mean there is a requirement for a high degree of maintenance.
Consequently, monitoring and diagnostics of wind turbines play an increasingly important role in the competitive operation of wind farms. The data generated by monitoring systems can be used to obtain models of wind turbines operating under different conditions, and hence predict output signals based on known inputs. By comparing output data obtained from operational turbines with those
predicted by the models, it is possible to detect changes that may be due to the presence of faults.
This paper discusses model-based condition monitoring methods for wind turbines, in which the relationships between measured variables are modelled using linear models and artificial neural networks identified from data
acquired from operational turbines. The relationships between variables are also modelled using non-linear state dependent ‘pseudo’ transfer functions. Although these state dependent parameter models have been used extensively as
the basis of non-linear controllers, the research described here represents the first occasion for which they have been employed for a condition monitoring system. It is found that artificial neural network-based models outperform state
dependent parameter models; however, the computing power required for the latter is considerably less. Finally, the monitoring data are used to develop adaptive threshold rules for critical output signals, forming the basis of an early