In this article, a novel harmonic subspace signature-based method is proposed for real-time detection and localization of inter-turn short circuit fault (ITSCF) in dual three-phase permanent magnet synchronous machines (DTP-PMSMs) which are driven by the vector space decomposition (VSD) control scheme. The method is independent of additional equipment, machine parameters, and prerecorded faulty data. Based on VSD theory, the ITSCF model of DTP-PMSMs is derived first. It reveals that the ITSCF can introduce an extra fundamental pulsating voltage vector to harmonic subspace, which has diverse locations in harmonic subspace according to the different faulty phases. Then, with different VSD transformation orientations, the ITSCF can be detected and located by comprehensively utilizing the fundamental components of harmonic subspace currents and reference voltages. Finally, the experimental results are given to verify the effectiveness of the proposed method.

Dual three-phase permanent magnet (DTPPM) machines are suitable for applications with high power density and high reliability requirements. However, uneven demagnetization fault (UDF) can occur due to harsh working conditions. This article proposes a non-invasive, online detection method of UDF in DTPPM machines based on the electrical signal difference between two sets of three-phase windings, which can be more conducive to UDF diagnosis with higher sensitivity. Firstly, the influence of dual three-phase winding arrangements on phase permanent magnet (PM) flux linkage is investigated in DTPPM machines under UDF. It shows that the fault characteristic harmonics appear in fractional-slot DTPPM machines and integral-slot DTPPM machines where a single set of three-phase windings is rotationally asymmetric. Further, according to the three-phase relationship of harmonic vectors, it is analyzed whether the fault characteristic harmonics are reflected in stator currents or zero-sequence voltage components (ZSVCs). Notably, taking advantage of the structural feature in DTPPM machines, the proposed method obtains higher sensitivity of UDF detection, as the amplitudes of fault characteristic harmonics in electrical signal difference are larger than that in stator current or ZSVC in a single set of three-phase windings. Finally, experiments are conducted to validate the effectiveness of the proposed method.

Eccentricity is a critical fault in permanent magnet synchronous machines (PMSMs), and can be generally classified into static eccentricity (SE), dynamic eccentricity (DE) and mixed eccentricity (ME). In this article, the fault diagnosis method based on zero-sequence voltage component (ZSVC) is proposed for detecting eccentricity in PMSMs with rotational asymmetric winding configurations. Different types of eccentricity can be classified with different fault characteristic harmonics in the measured ZSVC, which is extracted by Fast Fourier Transform (FFT). Specifically, the fundamental component and sideband harmonics appear under SE and DE, respectively. Both the fault characteristic components show up in case of ME. Notably, the ZSVC difference between two sets of three-phase windings obtains higher sensitivity for detection in dual three-phase PMSMs with rotational asymmetric winding configurations. Furthermore, the influence of eccentricity condition, load condition as well as other faults is investigated. Especially, the effect of d-axis current on fault indicators is revealed from armature flux linkage and PM flux linkage due to different degree of core saturation under eccentricity fault. Finally, the experiment is carried out on 10-pole/9-slot and 10-pole/12-slot surface-mounted single three-phase PMSMs, and the latter one is also connected as different dual three-phase winding configurations for verification.

Multiple synchronous reference frame (MSRF) current control scheme is frequently utilized in permanent-magnet synchronous machine (PMSM) drive systems for harmonic current suppression. However, low-pass filters (LPFs) with relatively low cutoff frequencies are commonly used to separate the fundamental and the major current harmonic components, which can introduce significant delays in the control loop, degrading the system performance. To address this issue, this article proposes a rapid and efficient harmonic current suppression scheme for PMSMs by incorporating a fast harmonic extraction method based on numerical calculation. For the separation of fundamental, 5t^h and 7^th harmonic currents, the proposed method only requires the current data from three sampling moments for processing, essentially constituting a time-shifting operation. Then the current with different orders can be decoupled and transformed into different harmonic planes, and independent closed-loop control can be realized. Compared with the existing MSRF-based current harmonic regulation methods, the proposed scheme avoids the use of any digital filter and does not require a large amount of data storage. This approach mitigates delay effects, enhancing the dynamic performance of harmonic current regulation, and offers easy implementation. Finally, comparative experimental validation is conducted to validate the effectiveness of the proposed scheme.

Multimodule permanent magnet synchronous machines (PMSMs) are identified as a better choice for superior fault-tolerant ability. However, as one of the most common faults, eccentricity makes the performance of PMSMs deteriorate. Without timely diagnosis and maintenance, negative effects caused by eccentricity further worsen the fault severity, even resulting in catastrophic accidents. In this article, the inherent rotational asymmetry of partition stator in multimodule PMSMs is utilized in the proposed zero-sequence voltage component (ZSVC)-based eccentricity fault diagnosis method with advantages of noninvasiveness, higher sensitivity, and the capacity to differentiate eccentricity types. Specifically, different from the previous ZSVC-based method monitoring multiple harmonics, the proposed method achieves that only the dominant 3rd ZSVC harmonic is extracted by short-time Fourier transform (STFT) in each unit of multimodule PMSMs. The variation of 3rd harmonic amplitude with rotor position shows asymmetry feature under static eccentricity (SE), fluctuation feature at mechanical rotating frequency under dynamic eccentricity (DE), and the above features under mixed eccentricity (ME). Then, the fault indicators are defined to detect the different eccentricity types. Furthermore, the influence factors are discussed exhaustively. Finally, experiments are implemented on an 8-pole/48-slot 4-module PMSM with SE shields and DE rotor to verify the effectiveness of the proposed approach.

Demagnetization represents a critical reliability challenge for permanent magnet synchronous machines (PMSMs) operating in demanding industrial environments. Nevertheless, its role in the generation of harmonic currents, as well as the development of effective suppression strategies for these harmonics, have remained insufficiently explored. This paper systematically analyzes the mechanisms by which demagnetization, especially non-uniform demagnetization, induces complex and diverse interharmonics (also known as fractional-order harmonics) in PMSM drives. A detailed mathematical model is developed, and finite element analysis is employed to quantify the changes in permanent magnet flux under various demagnetization scenarios. To address the challenge of suppressing these unpredictable harmonic disturbances, an online interharmonic compensation method based on a radial basis function neural network (RBFNN) is developed. Unlike conventional approaches, this method requires no prior knowledge of harmonic orders and can simultaneously mitigate multiple interharmonic components in real time. Experimental validation on a 10-pole/9-slot surface-mounted PMSM demonstrates the effectiveness and robustness of the proposed strategy under two representative demagnetization conditions, achieving effective suppression of demagnetization-induced harmonic currents.