A Closed-Loop PMSM Sensorless Control Based on the Machine Acoustic Noise

Abstract

In this paper, a novel sensorless control for Permanent Magnet Synchronous Machine (PMSM) using its radiated acoustic noise is proposed in which a highfrequency voltage signal is superimposed upon the fundamental excitation and the machine response to such excitation is studied. Firstly, the analytical equations describing the field formation within the airgap leading to the generation of radial forces and acoustic noise are extensively investigated. In the next step, resulting harmonics in the acoustic noise waveform caused by the injected signals are predicted. Then it is experimentally verified that, depending on the machine structure, there will be some distinctive sidebands around the injected frequency directly related to the fundamental synchronous frequency. Acoustic noise waveform is highly distorted and sensitive to the environment, so a proposed 4-stage signal processing algorithm containing two self-adaptive Band-pass Filters (BPF) are used to safeguard the signal against unwanted distortions and flawlessly extract the rotor position information. Robustness and effectivity of this method is verified experimentally in various operating conditions.

A Lean Approach to Zero and Low-Speed Sensorless Control of PMaSynRMs

“…Compared to the literature this is an improvement. Reference [13] shows a maximum position error of 0.606 at 75 rpm with a 1.6 kW PMSM while [14] shows a 0.573 degree error under no-load at 300 rpm and 11.459 degree under 0.14 pu load at 200 rpm. Also under 200% overload, [15] shows 1.318 degree error at 400 rpm for a 1 kW PMSM.…”

“…Considering the high-frequency nature of the injected voltages and low-speed operating region, the resistance and motor speed-dependent terms may be omitted, leaving only the highfrequency inductances [13], [14]. I.…”

“…The position information might be obtained directly from the remaining currents using an arctangent. However, the phase-locked loop (PLL) method provides a smoother estimation [13]. The PLL structure is a closed-loop estimator with a PI regulator for the position estimation as depicted in Fig.…”

“…In addition, both the estimated position and speed ripples increase with increasing load as well as during speed transients [11], [12]. Peak-to-peak estimation error also dramatically increases at higher speeds in [13] where HFRVI is used. Fluctuations in both speed and position at low speeds are attributed to the high inertia of the shaft.…”

Sensorless Control Strategy for Interior Permanent Magnet Synchronous Motors in the Full-Speed Section

“…Low-speed rotor position detection technology usually uses closed-loop control algorithms, i.e., high-frequency signal injection, which works by applying a certain frequency of voltage or current signals to a certain phase of the motor windings, relying on the motor's convex characteristics of the associated high-frequency current or voltage signals that contain position information and then digitally processing the signals to acquire position information. From the different types and forms of injected signals, high-frequency signal injection can be categorized into rotating high-frequency voltage injection [19][20][21] and pulse high-frequency voltage injection [22][23][24][25]. Although this method is simple, efficient and stable, the signal processing process is complicated.…”