Xingang Wang

Multichannel vibration from a planetary gearbox exhibits operating-condition-dependent interactions across both time scale and lag. Static undirected graphs cannot represent ordered temporal dependence, while repeated graph aggregation can attenuate localized impacts. This paper presents PG-MTCDGNet, a physics-grounded multi-scale time-lagged dynamic graph network for structured evidence reporting in planetary-gearbox fault diagnosis. Phase-synchronous X/Y vibration, transient residuals, Hilbert envelopes, cross-channel coupling, and encoder micro-speed are represented as eight physical nodes. Sparse directed graphs are induced using the angular lags of one, four, and 16 samples from lagged predictive associations and are regularized by label-independent mesh and modulation priors. A time-lagged graph branch is fused with an unsmoothed local-evidence branch, and the complete model is trained with a fold-internal noise-consistency curriculum spanning −6 to 10 dB. The WT dataset comprises 80 continuous raw files and 5120 windows from five health states, two installations, and eight speeds. Under raw-file-isolated, class-installation-speed-balanced five-fold validation, PG-MTCDGNet achieved 99.61% ± 0.33% accuracy, 99.63% ± 0.31% macro-F1, and 0.99999 macro-AUC. Accuracy at 0 dB feature noise was 96.78%. A depth-reduced variant, PG-MTCDGNet-Lite, separates comparable signal descriptors from heterogeneous physical tokens and uses a regularized linear readout. In exploratory installation and speed holdouts, its four-condition file-level accuracy was 88.125%, versus 86.250% for the matched MLP; likewise, macro-F1 was 88.044% versus 84.171%. Removing graph propagation tied its overall accuracy, so this improvement does not establish a unique graph contribution. The 80 recordings were reused during model development, and the results require confirmation on new independent recordings.

To address the challenge of high wind speed variability in wind farm planning, a small-sample-based spatio-temporal fusion and compression deep residual point prediction model, STiCDRS (Spatio-Temporal integration and Compression Deep Residual), is proposed. This model is designed to deeply explore the spatial and temporal characteristics within wind speed sequences to enhance the accuracy of point predictions. Initially, the spatio-temporal integration and compression deep residual network is employed to obtain point prediction results. Subsequently, an innovative hybrid model, STiCDRS-NKDE (STiCDRS-Nonparametric Kernel Density Estimation), is introduced to achieve interval predictions, thereby providing more reliable probabilistic forecasts of wind speed. The hyper-parameters of the model are optimized using Bayesian optimization, ensuring efficient and automated tuning. Finally, a case study involving wind speed forecasting at a wind farm in Inner Mongolia, China, is conducted, comparing the performance of the STiCDRS model with traditional models. Experimental results demonstrate that in comparison to other models, the proposed STiCDRS-NKDE model delivers superior point prediction accuracy, appropriate interval predictions, and reliable probabilistic forecasting outcomes, fully showcasing its significant potential in the domain of wind speed forecasting.

The emergence of AI companion robots is transforming the landscape of elderly care, offering numerous conveniences to senior citizens when their children are not around. This trend is particularly pertinent in ageing societies such as China. Against this backdrop, the present study aims to explore the acceptance of AI companion robots among the elderly from a user-centric perspective. By leveraging insights from existing studies in the literature, we identified three individual characteristic variables—technology optimism, innovativeness, and familiarity—to extend the Artificial Intelligence Device Use Acceptance (AIDUA) model. Subsequently, we developed a conceptual model which was empirically tested through structural equation modelling (SEM) analysis. Our dataset comprised responses from 452 elderly individuals in China. The results revealed that technology optimism and innovativeness were positively associated with performance expectancy and effort expectancy, whereas familiarity inversely predicted perceived risk. Furthermore, emotion was found to be positively influenced by performance expectancy and effort expectancy but negatively impacted by perceived risk. This research extends the AIDUA model within the context of AI companion robots by integrating individual characteristic variables. These findings offer valuable insights for the design and development of companion robots and enrich the domain of Human–Robot Interaction (HRI).

In light of the increasing demand for the reliability analysis of self-recovery products, with features of limited storage period and multistage degradation, a reliability evaluation model in whole life cycle is proposed. The degradation process comprises one storage phase and two working phases. On the basis of the idea of competitive failure, the shock process and the feature of self-recovery were introduced into the model. Furthermore, the problem of it being difficult to add variables to a reliability model is solved with the use of the Stieltjes integral. The influences of the parameters of model reliability are analyzed, and the results demonstrate that the new model could adequately describe the competing failure process. The model also exhibited certain feasibility and theoretical reference values.

The prediction of bearing remaining useful life (RUL) plays a pivotal role in ensuring the safe operation of machinery and reducing maintenance loss. Traditional prediction methods only consider the features of one domain or integrate the features of multiple domains into a one-dimensional sequence as the model input, which leads to some inaccuracy in prediction. In order to improve the prediction accuracy, a bearing RUL prediction method based on the parallel deep residual convolution neural network (P-ResNet), which is considered both time-domain features and time–frequency features, is proposed in this paper. Synchronous wavelet transform (SWT) is adopted to extract time–frequency features from original vibration signals. Both the time domain features and time–frequency domain features after dimension reduction by PCA are used as input to P-ResNet, which contains two series of parallel convolution operations to learn the time–frequency features and time-domain features, respectively, to ensure the comprehensiveness of information-bearing degradation. The residual layers were added to enhance the learning ability of time–frequency features. Kalman filter algorithm was used to smooth the prediction results. The IEEE PHM 2012 Data Challenge datasets were used as data sources for model training and prediction. Compared with the traditional convolutional neural network (CNN), the P-ResNet model maintains the synchronization of global and local information and has a stronger learning ability. The experiment data validate the effectiveness of the proposed method, and the comparison between the prosed methods and the others proves the superiority of the proposed method.

The degradation rates of many engineering systems are always positive in practical applications. In some cases, the degradation process is approximately linear, such as train wheel wear degradation and an increase in the operating current of laser devices. In this paper we introduce a degradation modeling and reliability estimation method based on the Wiener process with a transmuted-truncated normal distribution (TTND). The TTND is employed to characterize unit-to-unit variability due to its flexibility in capturing symmetry and the asymmetrical behavior of the drift variable of the Wiener process. A Wiener process with TTND is applied to model the degradation processes of the deteriorating systems in two cases. In the first case, we assume that the degradation process is unaffected by measurement uncertainties, whereas in the second case, the measurement uncertainties are considered. The exact and explicit expressions of PDFs and the reliability functions are derived based on the concept of first hitting time. The Gibbs sampling technique and the Metropolis–Hastings algorithm are employed to obtain the Bayesian estimates of the model’s parameters. The efficiency and the feasibility of the presented approach are demonstrated through numerical examples, with Monte Carlo simulation and a practical application with laser degradation data. Furthermore, deviance information criteria (DIC) are used to compare the proposed model and some existing models. The result indicated that the proposed approach provided better reliability estimation results.

The correlation between contact fatigue failure and wear failure of rolling bearing is analyzed, and the reliability model of rolling bearing based on multi-correlation failure mode is established. Based on the improved linear fatigue cumulative damage theory and wear theory, the limit state equations of two failure modes are established, and the correlation of state functions of contact fatigue and wear failure modes is described by the mixed Copula function. The random sample data are generated by Monte Carlo method, and the unknown parameters are estimated by genetic algorithm based on the minimum deviation square criterion, then the dynamic reliability model of rolling bearing is established. The quadratic polynomial without cross-term is used as the implicit functional proxy model, then the Latin hypercube sampling method and least squares theory are employed to estimate parameters of mixed Copula function and finally, the reliability sensitivity analysis of multi-failure modes of bearing is presented by fourth-order moment estimation method. Taking a certain type of spherical roller bearing as an example, the bearing reliability analysis considering the correlation of failure modes is carried out. The results show that compared with the bearing reliability model based on the assumption of independent failure modes and the weakest link theory, the bearing reliability model based on multi-correlation failure is more consistent with the practical application.