Xinwei Zhao

While deep learning has advanced industrial fault diagnosis, its reliance on massive labeled datasets remains a major obstacle to real-world application. In response to the scarcity of labeled fault samples and ineffective utilization of abundant unlabeled data, this paper proposes a novel semi-supervised fault diagnosis method based on supervision-guided contrastive learning (SGCL). First, a supervision-guided mechanism is introduced to leverage a small number of labeled samples to construct semantic class centers. This explicitly guides unlabeled data toward their respective categories, thereby alleviating the mismatch between different optimization objectives and enhancing intra-class compactness and inter-class separability. Second, a consistency regularization strategy is introduced to improve the model’s representation capability via pseudo-label constraints, enabling unlabeled representations to directly participate in updating the classifier parameters. By jointly optimizing these mechanisms, the proposed framework seamlessly translates raw unlabeled data into highly discriminative diagnostic features. Comprehensive experiments on a self-built dataset and the Paderborn dataset demonstrate that SGCL achieves diagnostic accuracies of 91.74% and 91.42%, respectively, under a 5% label ratio. This represents an average improvement of approximately 8%-10% over compared generative and contrastive semi-supervised methods, verifying the effectiveness of the proposed approach in addressing fault diagnosis tasks under severely limited label conditions.

In industrial operational environments, rolling bearing vibration signals not only contain fault related periodic impulse components, but also interference noises. The coupling of these signals attenuates the fault characteristics, particularly in cases of compound faults, significantly affecting the diagnostic accuracy of bearings. This paper proposes a rolling bearing fault separation and compound diagnosis method using a sparse Bayesian framework with adaptive prior knowledge. Firstly, the proposed method segments signals into a finite number of modes, employing filter banks to adaptively select decomposition modes with a new compound fault indicator, which reduces noise interference and eliminates irrelevant components while ensuring the preservation of fault-related information. Secondly, by employing an interpretable strategy enables fault frequency estimation, eliminating errors from empirical parameter settings and avoiding the computational complexity of parameter optimization. The feature frequencies are estimated based on the envelope harmonic product spectrum and used as a prior knowledge of sparse Bayesian learning. Finally, to avoid the restriction of current feature extraction methods that primarily focus on the selection and extraction of a single demodulation frequency band, a sparse Bayesian probabilistic model is particularly designed for compound fault diagnosis. Simulation and experimental results show that the proposed method can effectively extract and separate the fault features of each individual modes and realize the diagnosis of rolling bearings.

The widespread use of flip-chip technology in the field of microelectronics packaging makes defect detection technology face great challenges, which requires the development of detection technology with less manual intervention, a lightweight network architecture, and high precision. Aiming at the dual-stability of the neural architecture search (NAS), and explicitly associating the network architecture searched by NAS with the architecture features of vibration signals. This study proposes a dual-convergence sparse feature extractor (DSFE) for visualization vibration signals architecture feature searching, which attempts to learn a representation whose architecture feature can uniquely represent the intrinsic information of the target signal. DSFE analyzed and summarized three deficiencies of gradient-based NAS, namely, “huge GPU memory consumption,” “high collapse probability,” and “rigid number of node precursor operations.” The corresponding solution modules are proposed in turn, which are “primary and secondary search spaces,” “skip_connect coefficient modification,” and “dynamic sparse selection of precursor operations.” The effectiveness of DSFE is verified with flip-chip vibration signals excited by the air-coupled ultrasonic wave. By comprehensive comparative experiments with fixed-structure networks and gradient-based NAS methods, it is proved that DSFE cannot only achieve dual-stability of detection precision and architecture searched, but also ensure less resource consumption and multifarious architecture, which can broaden the application scope of DSFE in practical engineering.

Most accidents of centrifugal compressors are caused by fluid pulsation or unsteady fluid excitation. Rotating stall, as an unstable flow phenomenon in the compressor, is a difficult point in the field of fluid machinery research. In this paper, A stack denoising kernel autoencoder neural network method is proposed to study the early warning of rotating stall in a centrifugal compressor. By collecting the pressure pulsation signals of the centrifugal compressor under different flow rates in engineering practice, a double hidden layer sparse denoising autoencoder neural network is constructed. According to the output labels of the network, it can be judged whether the rotation stall occurs. At the same time, the Gaussian kernel is used to optimize the loss function of the whole neural network to improve the signal feature learning ability of the network. From the experimental results, it can be seen that the flow state of the centrifugal compressor is accurately judged, and the rotation stall early warning of the centrifugal compressor at different speeds is realized, which lays a foundation for the research of intelligent operation and maintenance of the centrifugal compressor.

Understanding of the fluid–structure interaction phenomena in centrifugal compressors is essential for the impeller structural integrity design. To enhance the understanding of blade vibration in a realistic flow environment, a single-stage centrifugal compressor representative of industrial architecture has been investigated experimentally. For deterministic synchronous vibrations, the response amplitudes at different operating lines are measured and compared. The fundamental relations between flow excitation and rotor modal resonance are established by pairwise strain gages from experimental perspective. Compared with the third mode, the impeller first bending mode shows traveling wave response characteristic only within the frequency band where two adjacent blades have the same resonant frequency. Besides, the impeller encounters unexpected nonsynchronous vibrations when operating near the flow instability boundary. Speed ramp tests show that the stall cell propagating speed increases along with the rotor rotational speed without changing the cell count. Response amplification is further measured when throttling the compressor into stall. Experimental findings point toward vaned diffuser rotating stall and the corresponding propagating pressure waves in circumference leads to the impeller vibration. The aerodynamic asymmetry of stall cells increases the possibility of aeroelastic coupling with the blade modes. These results contribute to an in-depth understanding of the aeroelastic phenomena in industrial centrifugal compressors. The observed nonsynchronous vibration is important for the aeroelastic design and also of great interest for numerical predictions near stall.

We address the problem of the reconstruction of time-frequency characteristics of sparse multi-band signals by using the discrete multi-coset sampling (DMCS) model. In this article, the signal is characterized by complicated time-variable components. According to the feature of the short-time Fourier transform (STFT) analysis method, we obtain the rewritten matrix form of the discrete STFT. Then, an analysis method of the multi-coset sliding window (MCSW) is proposed, and the discrete multi-coset sampling sequence is locally windowed. The sparse signal reconstruction algorithm is used to obtain the optimal time-frequency reconstruction value of the original signal. Numerical simulations show the feasibility of this method. We simulate the effects of measurement noise, sampling rate, and time-frequency analysis parameters on the reconstruction accuracy. Our method can carry out time-frequency reconstruction for undersampled signals obtained from multi-coset sampling to ensure the time-varying feature of the original signals, which can well highlight the characteristics of signals. The method is of great significance to the research and development of the sub-Nyquist sampling technique.

The unshrouded impeller is widely used in industrial centrifugal compressors and normally operates at high tip speed and large volume flow. However, this type of impeller can be very sensitive to flow excitations such as IGV wake, and hence encounters the challenge of high dynamic stress. Due to the lack of experimental vibration data, this paper aims to enhance the understanding of the IGV preswirl effect. The real operating representative data from strain gauges is acquired during the experiment. The blade transient and quasi-steady response due to upstream IGV wake under different configurations are investigated and quantified. Results show that the blade response increases with larger positive regulation. And under specific operating conditions, the vibration of the blade is quite large, which is comparable with synchronize resonance. This increment is attributed to the aerodynamic loading change due to enhanced distortion of the inlet flow. Based on the current findings, accurate numerical prediction of the blade forced vibration for a large shift of inlet flow condition is also needed for more reliable operating of the impeller.

As the unique rotating component, the impeller is the core component of a nuclear reactor coolant pump (RCP), the dynamic properties of the impeller are critical for the safe operation of the whole reactor. The purpose of this study was to shed comprehensive light on the pressure pulsation and modal properties of a scaled RCP impeller via experimental and numerical methods. The numerical model was validated by an experiment connecting the pressure pulsation signals at the diffuser inlet, and a good agreement was obtained between the numerical and experimental results. Pressure pulsation acting on the impeller's blade is mainly dominated by the impeller rotating frequency, the vane passing frequency, and the double blade‐passing frequency, and the pressure pulsation acting on the blade's pressure surface is more intense than on the suction surface. The modal properties were obtained via the modal test and numerical methods with the impeller suspended as a free body in the air and submerged inside water. The reduction in the impeller natural frequencies was between 31.63% and 37.77% for the corresponding mode shape due to the added mass effect of the fluid. Based on the pressure pulsation characteristics acting on the impeller and the natural frequency of the impeller, it is considered there is no risk of resonance in the impeller. Finally, it is expected that the present work can provide scientific guidance to avoid hydraulic resonance in nuclear reactor coolant pumps.

Blade tip timing (BTT) technology is a research hotspot and can directly estimate the blade vibration, which is of great significance to blade fault diagnosis. However, the traditional once per revolution (OPR) method has greatly limited the application of this technology due to the installation problem of the sensor and low measurement accuracy. In order to overcome the disadvantages of the OPR method, this article proposes a shifting straight-line fitting (SSLF) method to accurately calculate the blade vibration. The principle of the SSLF method is to obtain the expected arrival time of the optimal fitting point by shifting the straight-line fitting (SLF), and the vibration of the blade can be obtained according to the time. In this article, first, a brief introduction of BTT technology and the OPR method is given. Then, theoretical analysis and experimental investigations are used to verify the effectiveness of the SSLF method. In the end, the SSLF method is applied to the large centrifugal compressor, which can be concluded that the SSLF method can be used for blade monitoring of the compressor based on the investigation of experiments.

Rotating stall limits the operating range and stability of the centrifugal compressor and has a significant impact on the lifetime of the impeller blade. This paper investigates the relationship between stall pressure wave and its induced non-synchronous blade vibration, which will be meaningful for stall resonance avoidance at the early design phase. A rotating disc under a time-space varying load condition is first modeled to understand the physics behind stall-induced vibration. Then, experimental work is conducted to verify the model and reveal the mechanism of stall cells evolution process within flow passage and how blade vibrates when suffering such aerodynamic load. The casing mounted pressure sensors are used to capture the low-frequency pressure wave. Strain gauges and tip timing sensors are utilized to monitor the blade vibration. Based on circumferentially distributed pressure sensors and stall parameters identification method, a five stall cells mode is found in this compressor test rig and successfully correlates with the blade non-synchronous vibration. Furthermore, with the help of tip timing measurement, all blades vibration is also evaluated under different operating mass flow rate. Analysis results verify that the proposed model can show the blade forced vibration under stall flow condition. The overall approach presented in this paper is also important for stall vibration and resonance free design with effective experimental verification.