Chenhao Zhao

High-precision smooth speed regulation in permanent magnet synchronous motors (PMSMs) is challenged by torque ripple, parameter mismatches, and load fluctuations. Considering the uncertain nature of these disturbances, neural network (NN)-based methods provide superior applicability compared to internal model approaches due to their frequency-independent characteristics. However, most of these methods utilize feedforward NNs designed via asymptotic stability theory, which exhibit severe output distortions stemming from abrupt error surges and prolonged convergence times under varying operating conditions. To address these issues, this article proposes a robust speed control framework that integrates a recurrent NN (RNN) with fixed-time sliding-mode control (FTSMC), theoretically guaranteeing closed-loop fixed-time stability even under input saturation. Specifically, the RNN employs an internal recurrent loop to leverage historical context, thereby reducing the network output's susceptibility to instantaneous error spikes and alleviating transient distortions. Furthermore, incorporating FTSMC principles into the weight update law design accelerates the network's learning process, ensuring accurate approximation of periodic disturbances while shortening the transient period. Extensive experimental results validate the effectiveness of the proposed scheme across diverse operating scenarios.

Voltage-source-inverter (VSI) nonlinearity causes current distortion and degrades the accuracy of parameter identification and sensorless control in permanent magnet synchronous machine (PMSM) drives. This letter proposes a novel online estimation method for VSI nonlinearity voltage based on third-order generalized integrators (TOGIs). The impacts of limited current control bandwidth and phase misalignment on estimation accuracy are systematically investigated for the first time. To address these issues, the proposed method employs TOGIs to extract the sixth-order harmonic amplitudes from both the d-axis reference voltage u_d,ref and the dead-time coefficient D_d, and easily estimates the nonlinearity voltage through their ratio, thereby enabling more accurate estimation by effectively eliminating errors induced by limited current control bandwidth and phase misalignment. Experimental results validate the effectiveness and superiority of the proposed method.

Rotating high frequency voltage injection (HFVI) is a widely adopted technique for position sensorless control of interior permanent magnet synchronous motors (IPMSMs) at zero to low speed range. However, due to nonidealities in its implementation, conventional rotating HFVI methods suffer from phase shift effects that degrade position estimation accuracy and overall system performance. Most existing methods address only partial sources of phase shift, leaving residual uncompensated errors that can still reach approximately ten degrees. To achieve full compensation, this paper first systematically analyzes all potential sources of phase shift in rotating HFVI-based sensorless control. Based on the analysis, a comprehensive compensation strategy is proposed, containing resistance phase shift balancing using separate high-pass filters and residual phase shift cancellation using modified positive sequence current. By fully compensating the phase shift effects, the proposed method significantly enhances position estimation accuracy. Experimental results verify that the maximum estimation error is reduced to less than three degrees under no-load conditions and about five degrees under loaded conditions, while introducing only 1 μs of additional computation time compared with the existing methods, demonstrating its effectiveness and superiority.

This article proposes a seamless sensorless control method for interior permanent magnet synchronous motor (IPMSM) drives across the full speed range. The method employs two generalized integral extended state observers to separately estimate the high-frequency (HF) and fundamental electromotive force (EMF) components under the IPMSM active flux model with $d$-axis HF voltage injection. Two identical quadratic models are subsequently applied to the HF and fundamental EMFs. Subtracting their outputs yields a quadratic EMF that exhibits a large dc bias at zero-to-low speeds and increases monotonically with speed, thereby ensuring observability across the entire operating range. Compared with conventional full-speed-range sensorless methods, the proposed approach provides a seamless solution without the need to switch between different strategies. In addition, the proposed scheme is considerably simpler to implement than existing approaches, further enhancing its practicality for real-world applications. Comprehensive experimental results confirm the effectiveness of the proposed method.

Fault-tolerant ability of dual three-phase permanent magnet synchronous motors (PMSMs) is crucial for safety-critical applications. When a single open-phase fault occurs, the current references in conventional methods must be adjusted for avoiding the generation of torque ripple. Meanwhile, their design should balance the tradeoff between the torque operation range (TOR) and the copper loss. However, these postfault current references still need to be retailored under different open-phase fault locations; otherwise, more severe system damage may occur. Aiming to improve the robustness of the control algorithm, this article has proposed an online postfault strategy that can achieve minimum copper loss while maximizing the TOR for dual three-phase PMSMs with sinusoidal back-EMF. Unlike existing literature, the current reference in the harmonic subspace can be uniformly determined by the intermediate variable of phase shift and amplitude ratio, independent of fault scenarios. Additionally, the required fault identification is reduced from locating a specific phase to identifying one set of three-phase windings. The postfault performance of minimum copper loss can be achieved during the full TOR. Furthermore, the implementation of proposed algorithm does not require any optimization iterative process or look-up tables. Comprehensive experimental results have verified the effectiveness of the proposed method.

Smooth speed control is crucial for high-performance electric drives in electric aircraft applications. However, inevitable torque ripple sources, such as cogging torque and current sampling errors, can induce periodic speed oscillations. To solve this problem, the online-trained neural network (NN)-based method with arbitrary relationship approximation capability is proposed in this article to reject various harmonic disturbances of the speed loop. As the simplest structured NN, the adaptive linear neuron (ADALINE) can rapidly approximate and compensate for periodic disturbances with known harmonic orders, which however is not always the case in real applications. Therefore, roughly regarded as inserting an additional hidden layer into the ADALINE structure, the radial basis function NN directly using rotor position as the network's input is further proposed to effectively minimize speed ripple without the requirement of knowing disturbance frequencies, which greatly enhances the system's applicability and robustness. Various comparative experiments conducted on the dSPACE MicroLabBox-based permanent magnet synchronous motor platform validate the effectiveness of the proposed method.

With the growing demand for advanced control algorithms in permanent magnet synchronous motor (PMSM) speed regulation, active disturbance rejection control (ADRC) has garnered significant attention for its simplicity and effectiveness as an alternative to traditional proportional-integral (PI) controllers. However, two key challenges limit its broader application: the lack of an intuitive equivalence analysis that highlights the advantages of ADRC over PI control and the complexity in selecting appropriate extended state observer (ESO) structures within ADRC. To address these issues, this paper develops an equivalent model of ADRC based on the structure of a generalized PI controller, offering a clearer understanding of its operational principles. The results demonstrate the relationship between ADRC and generalized PI control while highlighting ADRC’s superior capabilities. Additionally, this paper constructs a generalized model that incorporates all ADRC observer configurations, including both high-order ESO (HESO) and cascaded ESO (CESO), enabling a comprehensive analysis of ADRC with various observer structures and establishing equivalence relationships between them. The findings provide valuable insights into the efficacy and versatility of ADRC in PMSM speed regulation, supported by experimental validation on a test bench using the dSPACE DS1202 MicroLabBox.

In order to enhance the position estimation accuracy of sensorless control of permanent magnet synchronous motor (PMSM), this brief proposes an improved super-twisting observer (STO) combined with a novel position compensation phase-locked loop (PLL) scheme. In this sensorless technology, the improved STO is designed using linear and nonlinear correction terms to achieve precise reconstruction of back electrimotive force (EMF) and finite-time convergence. With the back-EMF accurately estimated, the position compensation PLL is presented to alleviate the position error introduced by the speed acceleration and deceleration ramps when extracting position and speed information. Ultimately, the experimental comparisons on a 0.75-kW PMSM drive reveal that the proposed sensorless scheme has superior position estimation accuracy in both speed dynamics and steady state compared with the conventional schemes.