Variational auto-encoders (VAE) have recently been successfully applied in the intelligent fault diagnosis of rolling bearings due to its self-learning ability and robustness. However, the hyper-parameters of VAEs depend, to a significant extent, on artificial settings, which is regarded as a common and key problem in existing deep learning models. Additionally, its anti-noise capability may face a decline when VAE is used to analyze bearing vibration data under loud environmental noise. Therefore, in order to improve the anti-noise performance of the VAE model and adaptively select its parameters, this paper proposes an optimized stacked variational denoising autoencoder (OSVDAE) for the reliable fault diagnosis of bearings. Within the proposed method, a robust network, named variational denoising auto-encoder (VDAE), is, first, designed by integrating VAE and a denoising auto-encoder (DAE). Subsequently, a stacked variational denoising auto-encoder (SVDAE) architecture is constructed to extract the robust and discriminative latent fault features via stacking VDAE networks layer on layer, wherein the important parameters of the SVDAE model are automatically determined by employing a novel meta-heuristic intelligent optimizer known as the seagull optimization algorithm (SOA). Finally, the extracted latent features are imported into a softmax classifier to obtain the results of fault recognition in rolling bearings. Experiments are conducted to validate the effectiveness of the proposed method. The results of analysis indicate that the proposed method not only can achieve a high identification accuracy for different bearing health conditions, but also outperforms some representative deep learning methods.

Variational mode decomposition (VMD) with a non-recursive and narrow-band filtering nature is a promising time-frequency analysis tool, which can deal effectively with a non-stationary and complicated compound signal. Nevertheless, the factitious parameter setting in VMD is closely related to its decomposability. Moreover, VMD has a certain endpoint effect phenomenon. Hence, to overcome these drawbacks, this paper presents a novel time-frequency analysis algorithm termed as improved adaptive variational mode decomposition (IAVMD) for rotor fault diagnosis. First, a waveform matching extension is employed to preprocess the left and right boundaries of the raw compound signal instead of mirroring the extreme extension. Then, a grey wolf optimization (GWO) algorithm is employed to determine the inside parameters ( $\widehat{\alpha }$ , K) of VMD, where the minimization of the mean of weighted sparseness kurtosis (WSK) is regarded as the optimized target. Meanwhile, VMD with the optimized parameters is used to decompose the preprocessed signal into several mono-component signals. Finally, a Teager energy operator (TEO) with a favorable demodulation performance is conducted to efficiently estimate the instantaneous characteristics of each mono-component signal, which is aimed at obtaining the ultimate time-frequency representation (TFR). The efficacy of the presented approach is verified by applying the simulated data and experimental rotor vibration data. The results indicate that our approach can provide a precise diagnosis result, and it exhibits the patterns of time-varying frequency more explicitly than some existing congeneric methods do (e.g., local mean decomposition (LMD), empirical mode decomposition (EMD) and wavelet transform (WT) ).

In this paper, a synthetic sensorless starting scheme, which is based on pulse injection technology, is proposed for switched reluctance motor (SRM) drives to enhance the starting reliability. This scheme consists of a set of sensorless position estimation algorithms for different stages of SRM starting. In order to start the SRM from standstill stage smoothly, the initial driving phase is selected by pairwise comparing of the three-phase inductance. Then, to ensure reliable starting acceleration, a double inductance dynamic thresholds algorithm is developed for estimating the direct phase-driving signals. Meanwhile, to estimate the real-time rotor position, a position index pulse estimation algorithm, which is with phase lacking fault-tolerant capability, is developed based on an idle phase inductance threshold. By seamless switching between the above mentioned two algorithms, higher reliability of the sensorless starting control can be achieved. Furthermore, the performance of three pulse injection methods are compared, and the best one is then selected to reduce the influences of the high frequency pulse signals in the idle phases. The theoretical analysis and experimental results are provided to verify the feasibility of the proposed control scheme.

When rolling bearings have a local fault, the real bearing vibration signal related to the local fault is characterized by the properties of nonlinear and nonstationary. To extract the useful fault features from the collected nonlinear and nonstationary bearing vibration signals and improve diagnostic accuracy, this paper proposes a new bearing fault diagnosis method based on parameter adaptive variational mode extraction (PAVME) and multiscale envelope dispersion entropy (MEDE). Firstly, a new method hailed as parameter adaptive variational mode extraction (PAVME) is presented to process the collected original bearing vibration signal and obtain the frequency components related to bearing faults, where its two important parameters (i.e., the penalty factor and mode center-frequency) are automatically determined by whale optimization algorithm. Subsequently, based on the processed bearing vibration signal, an effective complexity evaluation approach named multiscale envelope dispersion entropy (MEDE) is calculated for conducting bearing fault feature extraction. Finally, the extracted fault features are fed into the k-nearest neighbor (KNN) to automatically identify different health conditions of rolling bearing. Case studies and contrastive analysis are performed to validate the effectiveness and superiority of the proposed method. Experimental results show that the proposed method can not only effectively extract bearing fault features, but also obtain a high identification accuracy for bearing fault patterns under single or variable speed.

In this article, an integrated power converter with dc bus voltage boosting capability is proposed for driving the brushless dc motors. In the topology, a bidirectional buck/boost dc–dc (BBD) converter can be integrated with the three-phase voltage-source inverter (TVSI) through a common three-switch leg. As compared with the traditional TVSI with a front-end BBD converter (TVSI-BBD), one power switch and its associate driving circuits in the BBD converter part can be saved. Thus, the integrated power converter has the benefits of reducing the converter size, lowering system costs, and enhancing power density. To decouple the dc bus voltage boosting control and the speed closed-loop control, a field-oriented control algorithm with space vector pulsewidth modulation technique is also proposed. The feasibility of the driving system is evaluated through systematic analysis and validated experimentally on a developed motor drive system prototype.

Due to the actual bearing working condition is changeable and complex, periodic impulse characteristics of bearing vibration data are easy to be contaminated by background noise and interference harmonics, which indicates that it is difficult to extract directly the fault feature information from the raw bearing vibration data. To address this issue, this paper proposes a bearing fault feature extraction method based on optimized singular spectrum decomposition (OSSD) and linear predictor (LP). Firstly, an LP is employed to preprocess the original bearing vibration data, which can eliminate effectively the influence of noise interferences in bearing vibration data on fault diagnosis. Secondly, the assisted indicator-based OSSD is presented to decompose the pretreated bearing vibration data into a series of singular spectrum components (SSCs), which can overcome the shortcomings of manual setting of mode number K in the original singular spectrum decomposition (SSD). Finally, the correlation kurtosis (CK) indicator is calculated to evaluate the richness of periodic impulses contained in each mode components, where the SSC component with the largest CK value is selected as the optimal SSC. Meanwhile, envelope demodulation analysis of the optimal SSC is conducted to extract bearing defect frequencies and judge bearing fault categories. The effectiveness of the proposed method is validated by using simulation analysis and two experimental cases. Experimental results show that the proposed method can extract effectively bearing defect frequencies. Moreover, fault feature extraction capability of the proposed method is better than that of the original SSD and four popular methods (i.e. parameter optimized variational mode decomposition, morphological filtering enhanced empirical wavelet transform, spectral kurtosis and singular value decomposition) involved in this paper.