In this dissertation, a modelling technique for motor driven system based on transduction matrix is introduced. Transduction matrix describes the relationship between input and output of a system in frequency domain, thus it provides better understanding of the system's condition. For a motor driven system running at constant frequency, the transduction matrices for induction motor, power transmission system and mechanical loading can be obtained from respective governing equations. Since transduction matrix represents the system's properties, condition monitoring can be attained by studying these transduction functions. In addition, transduction matrix facilitates the analysis on both impedance propagation and power flow in the motor driven system.Consequently, the electrical input impedance that consists of transduction function is utilized in condition monitoring of induction motor. Studying the frequency components of electrical input impedance allows characteristic frequencies of motor faults to be identified. Therefore, the proposed monitoring signature is used to monitor the change of signature due to broken rotor bar, abnormal air-gap eccentricity and bearing faults. By comparing with the results from conventional monitoring technique, impedance signature proved to have better sensitivity as fault signature is amplified. Furthermore, wavelet packet transform is utilized in order to carry out fault detection in time-frequency analysis, where the signal to noise ratio is improved and make the fault detection easier.