Hogir Rafiq proposes two approaches, the signal processing based condition monitoring approaches with applications to fault detection in gear systems, and application of deep mathematical and system theoretical methods to fault detection. The author develops the multivariate empirical mode decomposition (MEMD) algorithm to enhance the capability of extracting fault features and theoretical problems in nonlinear frequency analysis methods, respectively. The effectiveness has been demonstrated by an experimental study on a wind turbine gearbox test rig.