Servomechanism fault detection based on load torque observer and discrete wavelet transform

Most existing techniques which employ DWT (discrete wavelet transform) to detect servomechanism fault events are based on the analysis of measured motor current signals. However, the outcomes of these techniques tend to be unreliable when the motor suddenly accelerates or generates a very high output torque. Accordingly, to reduce the occurrence of false alarms, this paper develops a fault detection procedure based on the PICTO (PI-type closed-loop torque observer) structure and the DWT method. The PICTO structure is used to estimate the external load torque exerted on the servomechanisms. Utilizing the estimated data obtained from PICTO, a fault detection procedure is proposed which computes a "fault index" using DWT coefficients. This fault index enables the exact degree of the fault to be determined. The experimental results verify the effectiveness of the proposed fault detection approach.