This paper aims to classify local demagnetisation and inter-turn short-circuit (ITSC) on position sensorless permanent magnet synchronous motors (PMSM) in transient states based on external stray flux and learning classifier. Within the framework, four supervised machine learning tools were tested: ensemble decision tree (EDT), k-nearest neighbours (KNN), support vector machine (SVM), and feedforward neural network (FNN). All algorithms are trained on datasets from one operational profile but tested on other different operation profiles. Their input features or spectrograms are computed from resampled time-series data based on the estimated position of the rotor from one stray flux sensor through an optimisation problem. This eliminates the need for the position sensors, allowing for the fault classification of sensorless PMSM drives using only two external stray flux sensors alone. Both SVM and FNN algorithms could identify a single fault of the magnet defect with an accuracy higher than 95% in transient states. For mixed faults, the FNN-based algorithm could identify ITSC in parallel-strands stator winding and local partial demagnetisation with an accuracy of 87.1%.

Eccentricity and demagnetization fault of a four-pole 1.5 kW surface mounted permanentmagnet synchronous-generator (PMSG) were modelled by using time-discretised finite element analysis (FEA). Both fault types are caused by magnetic asymmetry in the generator. The faulty behaviour of a PMSG under transient operating condition is studied with FEA. Two search coils were wound around stator teeth on opposite sides of the rotor. The induced voltage from these coils will be equal in healthy case. A fault is detected when the induced voltages are non-identical. The simulation results revealed that the envelope of the induced search coil voltage had sinusoids during dynamic eccentricity and demagnetization. Finally, a novel fault scheme is proposed to detect the mentioned faults during transient state.

Conventional field reconstruction model (FRM) for electrical machines has proved its main strength in efficient computations of magnetic fields and forces in healthy permanent magnet synchronous machines (PMSM) or faulty machines in steady states. This study aims to develop a magnet library of different magnet defects and include inter-turn short-circuit (ITSC) in the FRM for PMSM. The developed FRM can model a combination fault between ITSC, and magnet defect in a PMSM in transient states. Within the framework, an 8-turn ITSC was modelled in both finite element analysis (FEA) and FRM, and then identified by the extended Park's vector approach. The air-gap magnetic field reproduced by the FRM shows a good agreement with the result from time-stepping FEA. The computation speed is over 1000 times faster than an equivalent time-stepping FEA. The suggested FRM allows for quickly understanding effect of faults in the rotor and stator on the air-gap magnetic flux density and identifying unique signatures for such defects.

This paper aims to improve quadratic time-frequency distributions to adapt condition monitoring of electrical machines in transient states. Short-Time Fourier transform (STFT) has been a baseline signal processing technique for detecting fault characteristic frequencies. However, limits of window sizes due to loss of frequency-or time-resolution, make it hard to capture rapid changes in frequencies. Within this study, Choi-Williams and Wigner-Ville distributions are proposed to effectively detect peaks at characteristic frequencies while still maintaining low computation time. The improved quadratic timefrequency distributions allow for generating spectrograms of a longer lasting data signal and capturing multi-component signals with a better separation of the components than STFT. Further, the time resolution of the spectrograms generated by the proposed method is not affected by the window size. The effectiveness of the proposed methods is numerically verified from the data of an in-house test setup.

This paper aims to improve quadratic time-frequency distributions to adapt condition monitoring of electrical machines in transient states. Short-Time Fourier transform (STFT) has been a baseline signal processing technique for detecting fault characteristic frequencies. However, limits of window sizes due to loss of frequency-or time-resolution, make it hard to capture rapid changes in frequencies. Within this study, Choi-Williams and Wigner-Ville distributions are proposed to effectively detect peaks at characteristic frequencies while still maintaining low computation time. The improved quadratic timefrequency distributions allow for generating spectrograms of a longer lasting data signal and capturing multi-component signals with a better separation of the components than STFT. Further, the time resolution of the spectrograms generated by the proposed method is not affected by the window size. The effectiveness of the proposed methods is numerically verified from the data of an in-house test setup.