Shaft misalignment detection using ANFIS for speed sensorless AC drive with inverter output filter

The aim of the paper is to present a diagnostic system for shaft misalignment detection. The diagnostic system is used in an adjustable speed sensorless induction motor (IM) drive with an inverter output filter. A nonlinear control algorithm and state observer were used in the motor control. Because the inverter output filter was installed in the drive, the control structure, as well as the estimation systems, were adequately changed. The adaptive neuro-fuzzy inference system (ANFIS) is used for shaft coupling fault detection. ANFIS is based on the analysis of the stator current, motor speed, and load torque processing. ANFIS uses only signals estimated from the state observers whereas observers calculate required variables only on the base of the inverter input voltage and two output current sensors. No additional special sensors are required. The proposed diagnostic system clearly indicates the shaft misalignment. The paper presents the model of the drive, control and estimation algorithms, as well as the diagnostic system. Whole drive and ANFIS fault indicator were verified by experiments for a 1.5 kW induction motor drive system.