Radial Displacement Sensorless Control of PMa-BSynRM Based on Square Wave Injection With Neural Network

A radial displacement sensorless control method of a permanent magnet assisted bearingless synchronous reluctance motor (PMa-BSynRM) based on square wave voltage injection (SWVI) is proposed to replace the sensor and enhance system reliability. First, the inductance expression of the PMa-BSynRM winding is derived based on the magnetic potential and magnetic permeability method when the rotor is eccentric. Then, a SWVI method of PMa-BSynRM is proposed. The high-frequency (HF) voltage signal is injected into the static coordinate system of the torque winding, and the induced current generated before and after the HF injection period is sampled on the suspension force winding. Through mathematical calculations, the radial displacement information of the rotor can be obtained. The phase-locked loop (PLL) is used as a rotor displacement observer to estimate rotor radial displacement. Considering that the observer based on fixed bandwidth design is difficult to achieve good control performance in a wide speed domain, a neural network algorithm is introduced to optimize the observer parameters online. Finally, the proposed method is validated on the experimental platform of the PMa-BSynRM. The experimental results demonstrate that the proposed radial displacement sensorless control method based on SWVI can achieve stable suspension operation with exceptional precision.