Rotating machinery fault diagnosis is essential for ensuring the reliability of industrial assets in IIoT-enabled manufacturing environments, where the high variability of operating conditions has driven the widespread adoption of multi-sensor fusion (MSF) to extract discriminative fault features. However, harsh IIoT environments frequently cause partial sensor failures or communication interruptions, under which conventional multi-sensor fusion systems often suffer significant performance degradation or even complete fusion failure. To address this issue, a Multi-task Multi-input Multi-scale Mixture-of-Experts (M4oE) framework is proposed, in which an independent diagnosis task is constructed for each sensor signal, ensuring that any individual sensor stream, representing the most extreme case of single-sensor availability, can independently perform diagnostic inference. First, a multi-scale mixture-of-experts feature extraction scheme is proposed, in which both the task-specific experts and each shared expert are implemented using the proposed Omni-scale Dilated Convolution Neural Network (OSD-CNN) architecture. Subsequently, task-specific features and shared features are integrated through multi-level feature fusion and fed into task-specific decoders to generate diagnostic results. Since the diagnostic outputs obtained from each sensor have the same identification framework and independent evidence sources, M4oE further employs Dempster-Shafer (D-S) decision-level fusion to enhance the reliability of the overall diagnostic system. Finally, comprehensive evaluations are conducted on three different types of rotating machinery datasets, including pumps, rolling mills, and bogies, to verify the accuracy, robustness, and scalability of the proposed M4oE framework.

Wind power gains more and more attention from all over the world as a clean and renewable energy resource, and accurate prediction of wind speed has become a hot issue. This paper presents a novel incremental forecaster based on phase space reconstruction and broad learning network (BLN) for short-term prediction. First, time delay and embedding dimension which take an essential part in phase space reconstruction are determined by the C–C algorithm. Then, these optimal parameters are input to the BLN trained incrementally. Afterward, forecasting values are given by the output layer of BLN. Data collected from a wind farm is adopted for verifying the efficacy of this proposed model. Furthermore, five commonly used assessment indicators are applied to evaluate predictive performance of different models. Results show that the proposed model has the smallest prediction error, which performs better than the other models at one-step to three-step ahead forecasting, and this strength is attributed to address the problem of local optimum. Furthermore, the proposed model consumes less training time than the other models. Therefore, the proposed model tends to be promising for wind speed prediction of the big data era.

Playing a vital role in keeping gearbox working reliably and safely, smart fault diagnosis (FD) technology has attracted much attention in recent years. However, in practical industrial applications, owing to the imbalance of healthy state data and fault state data and various unpredictable compound fault modes, it is still extremely challenging to fulfill the high-accuracy and effective FD of gearbox based on existing intelligent diagnostic models. In this paper, a new method is proposed to tackle these problems by integrating an improved deep residual network (IDRN) and wavelet transform (WT). The proposed method mainly contains two parts: In the first part, the vibration signals are transformed into images by WT. In the second part, the IDRN is applied to realize an accurate FD of gearbox. Compared with the original deep residual network, the main differences of IDRN are as follows: first of all, a new loss function is used to replace the commonly used logistic loss function by adding a class-balanced re-weighting term and combining with multi-label classification. Then, attention modules focusing operations on specific regions and enhancing the features of some regions are combined with residual blocks. Finally, a spatial transformer module is inserted into the original deep residual network to scale images and clip the region of interest. Two trials are conducted to validate the effectiveness of WT-IDRN method. The experimental consequences demonstrate that the WT-IDRN method has more excellent performance than the existing intelligent FD method in accuracy and generalization ability.

Transfer learning of multi-source domain adaptation seems a promising way for fault diagnosis of roller element bearings under variable working conditions. Data imbalance affects the performance of multi-source domain adaptation greatly and is expected to be solved by GAN. However, GAN-based transfer learning diagnosis models suffer pattern collapse and training instability, leading to unsatisfying diagnosis results in practical engineering. This paper proposes a denoising diffusion multi-source domain adaptation model (DDMDA). The proposed model uses diffusion denoising, which has better performance and is simpler to train than GAN, to generate shifted source domains for solving the data imbalance problem. A new noise prediction structure in diffusion denoising named Utrans-net, is constructed to restore the data distribution in the shifted source domain. Also, a multiple-domain discriminator structure is designed to extract features from multiple source domains to solve the issue of variable working conditions. Advanced models are used in this paper to compare with the proposed model for validation. Experimental demonstrations show that the proposed model is superior to the comparison models with satisfying performance.

Condition monitoring of machinery has entered the big data era, while the existence of dirty data reduces the quality of the whole data. In order to recognize the dirty data included in machinery monitoring data, a new method is proposed in this paper. First, a feature named sampled power index (SPI) is designed to transform the dirty data recognition issue into the outlier recognition. Then the windowing technique, the difference operation and the logarithm transform are introduced to reduce the feature tendency and the feature volatility. Next, auto regression-generalized autoregressive conditional heteroskedasticity (AR-GARCH) model is applied to regress the feature series and produce the crippled local means and local volatilities. Finally, the features are normalized and the 3σ criterion is applied to recognize the dirty data. The performance and the feasibility of this proposed method are evaluated by a simulation and an experiment. The results validate the effectiveness of the proposed method.