Wind turbine drivetrain vibrations are characterised by very complex dynamics originating from the convolution of deterministic and stochastic excitation sources moving through the structural dynamics of the wind turbine. In this work we postulate Linear Parameter Varying Vector AutoRegressive (LPV-VAR) models to represent those signals. To deal with the complexity of these models during the identification procedure, we devise an algorithm, based on the QR decomposition of the regression matrix, to accelerate the model identification procedure. The proposed methods are demonstrated on data from a wind turbine drivetrain simulator.