Condition Monitoring of Wind-Power Units Using the H-Infinity Kalman Filter

The article proposes H-infinity Kalman Filtering for condition monitoring of a wind-power generation system comprising a two-mass drivetrain and a DFIG. The H-infinity Kalman Filter is a robust state estimator which represents the fault-free functioning of the wind-power generation system. However, this filter is primarily designed for linear systems and to enable its use in the case of the nonlinear dynamics of the wind-turbine and DFIG, the state-space model of this power system undergoes first approximate linearization through Taylor series expansion and the computation of Jacobian matrices. By subtracting the outputs estimates provided by the H-infinity Kalman Filter from the real outputs of the power unit the residuals' sequence is generated. It is shown that the square of the sum of the residuals' vector, weighted by the inverse of the associated covariance matrix, forms a stochastic variable which follows the χ²-distribution. Next, by using the properties of the aforementioned distribution, confidence intervals are defined which allow for concluding with a high certainty if the monitored power system has undergone a failure or if it has been subject to a cyberattack. Actually, as long as the previously noted stochastic variable falls within the bounds of the confidence interval it is concluded that the functioning of the wind-power unit remains normal. Otherwise, when the bounds of the confidence interval are exceeded the existence of a fault or cyberattack can be detected. Finally, by applying the statistical test into subspaces of the wind-power unit's state-space description, fault or cyberattack isolation can be achieved.