Radial Displacement Measurement Method in Bearingless Slice Motor Through Eddy Current Displacement Sensors Positioned on the Underside of the Rotor

Abstract

Bearingless slice motors (BELSM) employing active radial positioning control and passive stabilization in the axial and tilt directions are widely utilized in medical and industrial applications. Conventionally, the radial displacement of a BELSM is measured by arranging eddy current displacement (ECD) sensors either outside or inside the rotor to detect its radial movement. However, this sensor placement method either increases the rotor’s height because of an additional sensor target or occupies the inner space because of the sensor holder. To address this issue, we propose the mounting of ECD sensors on the underside of the inner edge of the rotor for radial displacement measurements. However, formulating the radial displacement from the sensor outputs is challenging because of complex sensor nonlinearities and the effects of axial movement and tilt. To resolve this problem, we propose the usage of seven ECD sensor placements and establish relationships between the sensor outputs and radial displacement. Initially, displacement estimation based on the polynomial fitting method is proposed, however, it results in unacceptable measurement errors owing to sensor calibration errors. Subsequently, a neural network (NN) method is proposed, which exhibits good linearity within $\pm 150 \mu \text{m}$ and a measurement bandwidth of up to 200 Hz. The NN is further implemented into the displacement feedback system, enabling magnetic suspension startup and rotation with a radial displacement measurement error controlled within $16 \mu \text{m}$ within 1700 rpm. Diverse and large datasets, coupled with an increased number of sensors can substantially enhance measurement accuracy.

Keywords

• Bearingless slice motor
• magnetic suspension
• eddy current displacement sensor
• radial displacement measurement
• neural network