Search

Copyright of Rare Metals is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The number of fault samples is much less than the normal samples in the actual operation of the train bearing, and the imbalanced characteristics of the fault data significantly decrease the performance of the diagnosis model. Therefore, a train bearing imbalanced fault diagnosis method (ECCR-MFFN) based on extended combined cleaning and resampling (ECCR) and the multi-scale feature fusion network (MFFN) is proposed. Firstly, the ECCR method is proposed, which adaptively determines the sampling area and provides rich fault information for the diagnostic model with high-quality synthesized samples. Then, MFFN is designed to obtain great feature extraction and classification results under imbalanced data conditions through feature extraction and fusion strategies of multi-branch different kernels. Finally, the superiority and effectiveness of the ECCR-MFFN under various data imbalance conditions are verified by comparative experiments on laboratory and public bearing datasets. The results demonstrate that the MFFN can effectively extract fault features under small imbalance rate (IBR) conditions and achieve ideal classification results. Compared with other data augmentation methods, the ECCR can synthesize samples with higher quality and has a more stable performance. Under the condition of IBR = 40:1, the accuracy of the ECCR-MFFN is 95.84% and 96.07%, which is significantly better than the comparison methods and offers a reliable method for dealing with data imbalance. [ABSTRACT FROM AUTHOR]

High-dimensional and unlabeled data collected from multi-sensor is a common scenario in practical production. The fault diagnosis and health stage (HS) division of bearing under different degradation processes is easily limited due to unlabeled and high-dimensional data. This work designs an intelligent fault diagnosis and HS division strategy for unlabeled and high-dimensional data. An adaptive tensor density peaks search (ATDPS) clustering algorithm is proposed for the HS division of rolling bearing. Moreover, to enhance the clustering performance, a novel neighborhood least square (NLS) technique is developed for feature space denoising, whose effectiveness and superiority are verified compared with the other feature space denoising techniques. The proposed strategies are subsequently applied to three benchmark cases and compared with other clustering methods. The experiment results demonstrate that the proposed strategy can reliably and accurately divide the different degradation stages depending on less prior knowledge. Furthermore, the presented HS division approach successfully monitors degradation with compound failure, showing potential for practical application. [ABSTRACT FROM AUTHOR]

As one of the vital components of trains, the condition of train bearings is closely related to the safe operation of trains. Traditional bearing fault diagnosis methods based on single sensors are incapable of extracting feature information fully, resulting in low fault diagnostic accuracy. To solve the above problem, a fault diagnosis method for train bearings based on multi-sensor data fusion and dual-scale residual network (MSDF-DSRNet) is proposed in this paper. Firstly, a multi-sensor data fusion method is designed to extract fault feature information comprehensively. The low-dimensional features embedded in the high-dimensional nonlinear space of multi-sensor data are extracted effectively and fused into a three-dimensional pixel matrix. Secondly, a novel intelligent diagnosis method is proposed based on the dual-scale residual network. Both in-depth and shallow features are learned on two scales, and the fault-related information in different spatial dimensions is captured, which improves the extraction ability of fusion features and effectively reduces time loss. Finally, the feasibility and effectiveness of the proposed method are verified by three experiments. The accuracy of the proposed method in the train traction motor bearing dataset reaches 99.70%, 99.75%, 99.85% and 99.85%, respectively. The results show that MSDF-DSRNet performs better in comprehensive fault diagnosis than other methods. [ABSTRACT FROM AUTHOR]

The positive- and negative-ion electrospray ionization (ESI) coupled with Fourier transform-ion cyclotron resonance mass spectrometry (FT-ICR MS) was employed to identify the chemical composition of heteroatomic compounds in four distillates of Fushun shale oil, and their catalytic cracking performance was investigated. There are nine classes of basic nitrogen compounds (BNCs) and eleven classes of non-basic heteroatomic compounds (NBHCs) in the different distillates. The dominant BNCs are mainly basic N1 class species. The dominant NBHCs are mainly acidic O2 and O1 class species in the 300–350 °C, 350–400 °C, and 400–450 °C distillates, while the neutral N1, N1O1 and N2 compounds become relatively abundant in the > 450 °C fraction. The basic N1 compounds and acidic O1 and O2 compounds are separated into different distillates by the degree of alkylation (different carbon number) but not by aromaticity (different double-bond equivalent values). The basic N1O1 and N2 class species and neutral N1 and N2 class species are separated into different distillates by the degrees of both alkylation and aromaticity. After the catalytic cracking of Fushun shale oil, the classes of BNCs in the liquid products remain unchanged, while the classes and relative abundances of NBHCs vary significantly. [ABSTRACT FROM AUTHOR]

Polysaccharides and ganoderic acids (GAs) are the major bioactive constituents of Ganoderma species. However, the commercialization of their production was limited by low yield in the submerged culture of Ganoderma despite improvement made in recent years. In this work, twelve Ganoderma strains were screened to efficiently produce polysaccharides and GAs, and Ganoderma lucidum 5.26 (GL 5.26) that had been never reported in fermentation process was found to be most efficient among the tested stains. Then, the fermentation medium was optimized for GL 5.26 by statistical method. Firstly, glucose and yeast extract were found to be the optimum carbon source and nitrogen source according to the single-factor tests. Ferric sulfate was found to have significant effect on GL 5.26 biomass production according to the results of Plackett-Burman design. The concentrations of glucose, yeast extract and ferric sulfate were further optimized by response surface methodology. The optimum medium composition was 55 g/L of glucose, 14 g/L of yeast extract, 0.3 g/L of ferric acid, with other medium components unchanged. The optimized medium was testified in the 10-L bioreactor, and the production of biomass, IPS, total GAs and GA-T enhanced by 85, 27, 49 and 93 %, respectively, compared to the initial medium. The fermentation process was scaled up to 300-L bioreactor; it showed good IPS (3.6 g/L) and GAs (670 mg/L) production. The biomass was 23.9 g/L in 300-L bioreactor, which was the highest biomass production in pilot scale. According to this study, the strain GL 5.26 showed good fermentation property by optimizing the medium. It might be a candidate industrial strain by further process optimization and scale-up study. [ABSTRACT FROM AUTHOR]

Aim: To discriminate 32 phenethyl-amines between antagonists and agonists, and predict the activities of these compounds. Methods: The support vector machine (SVM) is employed to investigate the structure-activity relationship (SAR)/quantitative structure-activity relationship (QSAR) of phenethyl-amines based on molecular descriptors. Results: By using the leave-one-out cross-validation (LOOCV) test, 1 optimal SAR and 2 optimal QSAR models for agonists and antagonists were attained. The accuracy of prediction for the classification of phenethyl-amines by using the LOOCV test is 91.67%, and the accuracy of prediction for teh classification of phenethyl-amines by using the independent test is 100%; the results are better than those of the Fisher, the artificial neural network (ANN), and the K-nearest neighbor models for this real world data. The RMSE (root mean square error) of antagonists' QSAR model is 0.5881, and the RMSE of agonists' QSAR model is 0.4779, which are better than those of the multiple linear regression, partial least squares, and ANN models for this real world data. Conclusion: The SVM can be used to investigate the SAR and QSAR of phenethyl-amines and could be a promising tool in the field of SAR/QSAR research. [ABSTRACT FROM AUTHOR]

In this work, support vector classification (SVC) algorithm was used to build structure-activity relationship (SAR) model of the 5-hydroxytryptamine type 3 (5-HT3) receptor antagonists with 26 compounds. In a benchmark test, SVC was compared with several techniques of machine learning currently used in the field. The prediction performance of the model was discussed on the basis of the leave-one-out cross-validation. The results show that the accuracy of prediction of SVC model was higher than those of back propagation artificial neural network (BP ANN), K-nearest neighbor (KNN) and Fisher methods. [ABSTRACT FROM AUTHOR]