This paper compares pulsating- and rotating-vector carrier-signal injection for self-sensing (or sensorless) control of permanent-magnet synchronous machines (PMSMs). The main focus of this paper is the evaluation of estimation errors due to the physical nonideal attributes of both the machine and the inverter. Initial-position and magnet-polarity estimations are analyzed as well. Transient response and signal processing for both techniques are discussed for completeness. The theoretical analysis is supported by experimental and simulation evidence obtained using interior PMSMs.

This paper proposes a method to estimate the magnet temperature in surface permanent-magnet machines using high-frequency carrier signal injection. The injection of a high-frequency signal, superimposed on the fundamental excitation, allows the estimation of the stator high-frequency impedance, which is a function of both the stator and rotor impedances. The temperature of the magnets is shown to have a significant weight on the overall stator high-frequency impedance, from which it can be estimated. The high-frequency carrier signal is injected intermittently in order to minimize potential adverse effects on the normal operation of the machine. This paper first explains the physics behind the magnet temperature dependence. Then, the principles of the method, as well as its practical implementation, are discussed. Experimental verification of the method is provided.

Islanding detection is of great importance for reliable operation of smart grids. Islanding-detection methods can be classified into three different groups, i.e., active, passive, and communication-based methods. Active methods inject a disturbing signal (typically a voltage) and analyze the system response (typically in the current). These techniques have a low nondetection zone (NDZ) but present the inconvenience inherent to injecting a disturbing signal. Passive methods monitor the grid condition from the grid variables. These techniques are easy to implement but present a large NDZ. Communication methods have the inconvenience of relying on communications, currently being of limited use. This paper proposes a new active islanding-detection method, based on the measurement of the grid high-frequency impedance by means of the injection of a high-frequency voltage. The advantages of the method are almost negligible adverse effects due to the injected high-frequency voltage and accurate and fast islanding detection, i.e., in the range of a few milliseconds. Furthermore, the estimated high-frequency impedance can be used for the adaptive control of the power converter.

The injection of a high-frequency signal in the stator via inverter has been shown to be a viable option to estimate the magnet temperature in permanent-magnet synchronous machines (PMSMs). The variation of the magnet resistance with temperature is reflected in the stator high-frequency resistance, which can be measured from the resulting current when a high-frequency voltage is injected. However, this method is sensitive to d- and q-axis inductance (L _d and L _q ) variations, as well as to the machine speed. In addition, it is only suitable for surface PMSMs (SPMSMs) and inadequate for interior PMSMs (IPMSMs). In this paper, the use of a pulsating high-frequency current injection in the d-axis of the machine for temperature estimation purposes is proposed. The proposed method will be shown to be insensitive to the speed, L _q , and L _d variations. Furthermore, it can be used with both SPMSMs and IPMSMs.

This paper analyzes the effects of magnetic saturation, including cross-saturation and secondary saliencies, on the saliency-based sensorless control of interior permanent magnet synchronous machines. These effects are mitigated by adaptively decoupling saturation-induced saliencies via a structured neural network. This paper includes the identification of the dominant saturation-induced components of the carrier-signal current interfering with the rotor-position-dependent component being tracked, characterization of these components, and implementation of a nonlinear adaptive saturation-induced-component structured neural network model to perform their decoupling.

Rotor eccentricity and local demagnetization in permanent magnet synchronous motors (PMSMs) increases unbalanced magnetic pull and motor vibration resulting in accelerated aging of motor components. If the asymmetry in the rotor remains undetected, it can increase in severity, and increase the risk of stator-rotor contact, which causes forced outage of the motor and driven process. Detection of PMSM rotor asymmetry currently relies on off-line testing and on-line vibration/current spectrum analysis. However, they are inconvenient or cannot provide reliable detection as they are influenced by other mechanical non-idealities in the motor or load. In this paper, the feasibility of using the signals from analog Hall-effect field sensors for detecting eccentricity and local demagnetization is investigated. It is shown that Hall sensors present in machines for motion control can be used for measuring the variation in the flux inside the motor due to rotor magnetic asymmetry with minimal or no hardware modifications. 3-dimensional (3D) finite element analysis (FEA) and experimental results performed on an interior PMSM (IPMSM) show that the proposed method can provide sensitive and reliable detection of dynamic/mixed eccentricity and local PM demagnetization.