Machinery generally operates under severe and complex conditions and therefore the monitoring signals acquired from machinery would inevitably be accompanied by various types of noise in the process of data acquisition. Noise would result in the instability of intelligent fault diagnosis and prediction models and decline their recognition and prediction accuracy. In stochastic resonance, however, noise is beneficial to weak signal detection and intelligent image classification, while the research on the benefits of noise in mechanical intelligent fault diagnosis is still rare. For this purpose, the benefits of noise to the intelligent fault diagnosis are studied in this paper by injecting different levels of Gaussian and uniform noise to intelligent fault diagnosis models and even their input data sets. Then, an intelligent fault diagnosis method enhanced by injecting moderate noise is proposed to improve the classification accuracy of those ones without noise injection. Finally, three experiments including hydraulic motors and two different motor bearings were performed to verify the proposed method. The experimental results show that the diagnosis accuracy of hydraulic motors and two different motor bearings after noise injection is 95%, 95.6% and 97.5% respectively, which is increased by 1.4%, 1.6% and 1.1% than those without noise injection. Comparing the experimental results by injecting two different types of noise, all of them have the same optimal noise level to achieve fairly high classification accuracy. In addition, it is found that the diagnosis accuracy by injecting Gaussian noise is higher than that by injecting uniform noise.

Aptly enabled by recent developments in additive manufacturing technology, the concept of functionally grading some cementitious composites to improve structural compression forms is warranted. In this work, existing concrete models available in Abaqus Finite Element (FE) packages are utilized to simulate the performance of some cementitious composites numerically and apply them to functional grading using the multi-layer approach. If yielding good agreement with the experimental results, two-layer and three-layer models case combinations are developed to study the role of layer position and volume. The optimal and sub-optimal performance of the multi-layer concrete configurations based on compressive strength and sustained strains are assessed. The results of the models suggest that layer volume and position influence the performance of multi-layer concrete. It is observed that when there exists a substantial difference in material strengths between the concrete mixes that make up the various layers of a functionally graded structure, the influence of position and of material volume are significant in a two-layer configuration. In contrast, in a three-layer configuration, layer position is of minimal effect, and volume has a significant effect only if two of the three layers are made from the same material. Thus, a multilayered design approach to compression structures can significantly improve strength and strain performance. Finally, application scenarios on some structural compression forms are shown, and their future trajectory is discussed.

At present, the condition monitoring of wind turbine gearbox mostly involves selecting monitoring indicators in advance and establishing normal behavior models based on deep learning. However, single physical monitoring indicator cannot fully characterize the operating status of the wind turbine gearbox. To address the above issues, a wind turbine gearbox condition monitoring method using convolutional neural network (CNN)-long and short term memory network (LSTM)-autoencoder (AE) enabled virtual indicators is proposed in this paper. This method first establishes a cleaning principle to preprocess the data of supervisory control and data acquisition (SCADA) system and screen out effective SCADA data further. A CNN-LSTM-AE monitoring model was trained using SCADA data during normal operating process, various characteristic parameters of the gearbox are integrated to construct the AI-enabled virtual indicators, and meanwhile warning thresholds was determined based on the probability density distribution of virtual indicators to analyze the operating status of the gearbox. Finally, the proposed method was validated using real SCADA data from two wind turbines in a cooperated wind farm with my group. Compared with a single physical indicator, the virtual indicator enables to detect gearbox early faults 5 d in advance, indicating that the proposed method can effectively alert wind turbine gearbox failures.

As a powerful feature extraction tool, a convolutional neural network (CNN) has strong adaptability for big data applications such as bearing fault diagnosis, whereas the classification performance is limited when the quality of raw signals is poor. In this paper, stochastic resonance (SR), which provides an advanced feature enhancement approach for weak signals with strong background noise, is introduced as a data pre-processing method for the CNN to improve its classification performance. First, a multiparameter adjusting bistable Duffing system that can achieve SR under large-parameter weak signals is introduced. A hybrid optimization algorithm (HOA) combining the genetic algorithm (GA) and the simulated annealing (SA) is proposed to adaptively obtain the optimized parameters and output signal-to-noise ratio (SNR) of the Duffing system. Therefore, the data optimization based on the multiparameter-adjusting SR of Duffing system can be realized. An SR-based mapping method is further proposed to convert the outputs of the Duffing system into grey images, which can be further processed by a normal CNN with batch normalization (BN) layers and dropout layers. After verifying the feasibility of the HOA in multiparameter optimization of the Duffing system, the bearing fault data set from the CWRU bearing data center was processed by the proposed fault enhancement classification and identification method. The research showed that the weak features of the bearing signals could be enhanced significantly through the adaptive multiparameter optimization of SR, and classification accuracies for 10 categories of bearing signals could achieve 100% and those for 20 categories could achieve more than 96.9%, which is better than other methods. The influences of the population number on the classification accuracies and calculation time were further studied, and the feature map and network visualization are presented. It was demonstrated that the proposed method can realize high-performance fault enhancement classification and identification.

The performance of bearing approximately satisfies ergodicity property, therefore, investigating it in the spatiotemporal domain is theoretically feasible. This paper pioneers the application of Empirical Orthogonal Functions (EOF) from meteorological analysis and forecasting to the intelligent maintenance field of rotating machinery. Firstly, a Hankel matrix is constructed by using a column vibration signal, then we use EOF for the Hankel matrix to gain a time matrix, a space matrix, and a contribution rate vector. Secondly, some indicators are constructed based on the time matrix, whereas a brand-new spatiotemporal indicator is designed based on contribution rate vector. By calculating the total contribution rate and selecting some vectors in time and space matrices, the reconstructed signal can be obtained, and then some other indicators are constructed by using the reconstructed signal. This paper addresses two main challenges: first, decreasing the effect of noise, thus improving the ability to detect early failure in complicated environment; second, separating vibration signals into mutually orthogonal quantities in time-space perspective. The proposed method is validated on two bearing datasets, and the results show that the proposed indicators can either reflect the early defect or assess the degradation trend. Among all indicators, the brand-new designed indicator in spatiotemporal domain achieves the two goals simultaneously: when the early failure occurs, the indicator indicates a huge change, and with the further degradation of the bearing health state it remains good monotonicity and trend. Therefore, the designed health indicators enrich the family of statistical health indicators for machinery.