A Novel Fault Diagnosis Method of High-Speed Train Based on Few-Shot Learning.

Ensuring the safe and stable operation of high-speed trains necessitates real-time monitoring and diagnostics of their suspension systems. While machine learning technology is widely employed for industrial equipment fault diagnosis, its effective application relies on the availability of a large dataset with annotated fault data for model training. However, in practice, the availability of informational data samples is often insufficient, with most of them being unlabeled. The challenge arises when traditional machine learning methods encounter a scarcity of training data, leading to overfitting due to limited information. To address this issue, this paper proposes a novel few-shot learning method for high-speed train fault diagnosis, incorporating sensor-perturbation injection and meta-confidence learning to improve detection accuracy. Experimental results demonstrate the superior performance of the proposed method, which introduces perturbations, compared to existing methods. The impact of perturbation effects and class numbers on fault detection is analyzed, confirming the effectiveness of our learning strategy.

Research on Performance Degradation Estimation of Key Components of High-Speed Train Bogie Based on Multi-Task Learning.

The safe and comfortable operation of high-speed trains has attracted extensive attention. With the operation of the train, the performance of high-speed train bogie components inevitably degrades and eventually leads to failures. At present, it is a common method to achieve performance degradation estimation of bogie components by processing high-speed train vibration signals and analyzing the information contained in the signals. In the face of complex signals, the usage of information theory, such as information entropy, to achieve performance degradation estimations is not satisfactory, and recent studies have more often used deep learning methods instead of traditional methods, such as information theory or signal processing, to obtain higher estimation accuracy. However, current research is more focused on the estimation for a certain component of the bogie and does not consider the bogie as a whole system to accomplish the performance degradation estimation task for several key components at the same time. In this paper, based on soft parameter sharing multi-task deep learning, a multi-task and multi-scale convolutional neural network is proposed to realize performance degradation state estimations of key components of a high-speed train bogie. Firstly, the structure takes into account the multi-scale characteristics of high-speed train vibration signals and uses a multi-scale convolution structure to better extract the key features of the signal. Secondly, considering that the vibration signal of high-speed trains contains the information of all components, the soft parameter sharing method is adopted to realize feature sharing in the depth structure and improve the utilization of information. The effectiveness and superiority of the structure proposed by the experiment is a feasible scheme for improving the performance degradation estimation of a high-speed train bogie.

Between-Class Adversarial Training for Improving Adversarial Robustness of Image Classification.

Deep neural networks (DNNs) have been known to be vulnerable to adversarial attacks. Adversarial training (AT) is, so far, the only method that can guarantee the robustness of DNNs to adversarial attacks. However, the robustness generalization accuracy gain of AT is still far lower than the standard generalization accuracy of an undefended model, and there is known to be a trade-off between the standard generalization accuracy and the robustness generalization accuracy of an adversarially trained model. In order to improve the robustness generalization and the standard generalization performance trade-off of AT, we propose a novel defense algorithm called Between-Class Adversarial Training (BCAT) that combines Between-Class learning (BC-learning) with standard AT. Specifically, BCAT mixes two adversarial examples from different classes and uses the mixed between-class adversarial examples to train a model instead of original adversarial examples during AT. We further propose BCAT+ which adopts a more powerful mixing method. BCAT and BCAT+ impose effective regularization on the feature distribution of adversarial examples to enlarge between-class distance, thus improving the robustness generalization and the standard generalization performance of AT. The proposed algorithms do not introduce any hyperparameters into standard AT; therefore, the process of hyperparameters searching can be avoided. We evaluate the proposed algorithms under both white-box attacks and black-box attacks using a spectrum of perturbation values on CIFAR-10, CIFAR-100, and SVHN datasets. The research findings indicate that our algorithms achieve better global robustness generalization performance than the state-of-the-art adversarial defense methods.

Immune Escape Adaptive Mutations in Hemagglutinin Are Responsible for the Antigenic Drift of Eurasian Avian-Like H1N1 Swine Influenza Viruses.

The continuous antigenic variation of influenza A viruses remains a major hurdle for vaccine selection; however, the molecular determinants and mechanisms of antigenic change remain largely unknown. In this study, two escape mutants were generated by serial passages of the Eurasian avian-like H1N1 swine influenza virus (EA H1N1 SIV) A/swine/Henan/11/2005 (HeN11) in the presence of two neutralizing monoclonal antibodies (mAbs) against the hemagglutinin (HA) protein, which were designated HeN11-2B6-P5 and HeN11-4C7-P8, respectively. The HeN11-2B6-P5 mutant simultaneously harbored the N190D and I230M substitutions in HA, whereas HeN11-4C7-P8 harbored the M269R substitution in HA (H3 numbering). The effects of each of these substitutions on viral antigenicity were determined by measuring the neutralization and hemagglutination inhibition (HI) titers with mAbs and polyclonal sera raised against the representative viruses. The results indicate that residues 190 and 269 are key determinants of viral antigenic variation. In particular, the N190D mutation had the greatest antigenic impact, as determined by the HI assay. Further studies showed that both HeN11-2B6-P5 and HeN11-4C7-P8 maintained the receptor-binding specificity of the parent virus, although the single mutation N190D decreased the binding affinity for the human-type receptor. The replicative ability in vitro of HeN11-2B6-P5 was increased, whereas that of HeN11-4C7-P8 was decreased. These findings extend our understanding of the antigenic evolution of influenza viruses under immune pressure and provide insights into the functional effects of amino acid substitutions near the receptor-binding site and the interplay among receptor binding, viral replication, and antigenic drift. IMPORTANCE The antigenic changes that occur continually in the evolution of influenza A viruses remain a great challenge for the effective control of disease outbreaks. Here, we identified three amino acid substitutions (at positions 190, 230, and 269) in the HA of EA H1N1 SIVs that determine viral antigenicity and result in escape from neutralizing monoclonal antibodies. All three of these substitutions have emerged in nature. Of note, residues 190 and 230 have synergistic effects on receptor binding and antigenicity. Our findings provide a better understanding of the functional effects of amino acid substitutions in HA and their consequences for the antigenic drift of influenza viruses.

Performance Degradation Estimation of High-Speed Train Bogie Based on 1D-ConvLSTM Time-Distributed Convolutional Neural Network.

High-speed train bogies are essential for the safety and comfort of train operation. The performance of the bogie usually degrades before it fails, so it is necessary to detect the performance degradation of a high-speed train bogie in advance. In this paper, with two key dampers on the bogie taken as experimental objects (lateral damper and yaw damper), a novel 1D-ConvLSTM time-distributed convolutional neural network (CLTD-CNN) is proposed to estimate the performance degradation of a high-speed train bogie. The proposed CLTD-CNN is an encoder-decoder structure. Specifically, the encoder part of the proposed structure consists of a time-distributed 1D-CNN module and a 1D-ConvLSTM. The decoder part consists of a 1D-ConvLSTM and a simple time-CNN with residual connections. In addition, an auxiliary training part is introduced into the structure to support CLTD-CNN in learning the performance degradation trend characteristic, and a special input format is designed for this structure. The whole structure is end-to-end and does not require expert knowledge or engineering experience. The effectiveness of the proposed CLTD-CNN is tested by the high-speed train CRH380A under different performance states. The experimental results demonstrate the superiority of CLTD-CNN. Compared to other methods, the estimation error of CLTD-CNN is the smallest.

A Single Amino Acid at Position 431 of the PB2 Protein Determines the Virulence of H1N1 Swine Influenza Viruses in Mice.

Genetic reassortments occurred continuously among multiple subtypes or genotypes of influenza viruses prevalent in pigs. Of note, some reassortant viruses bearing the internal genes of the 2009 pandemic H1N1 (2009/H1N1) virus sporadically caused human infection, which highlights their potential threats to human public health. In this study, we performed phylogenetic analysis on swine influenza viruses (SIVs) circulating in Liaoning Province, China. A total of 22 viruses, including 18 H1N1 and 4 H1N2 viruses, were isolated from 5,750 nasal swabs collected from pigs in slaughterhouses from 2014 to 2016. H1N1 viruses formed four genotypes, which included Eurasian avian-like H1N1 (EA H1N1) and double/triple reassortant H1N1 derived from EA H1N1, 2009/H1N1, and triple reassortant H1N2 (TR H1N2) viruses. H1N1 SIVs with different genotypes and even those within the same genotypes represented different pathogenicities in mice. We further characterized two naturally isolated H1N1 SIVs that had similar viral genomes but differed substantially in their virulence in mice and found that a single amino acid at position 431 in the basic polymerase 2 (PB2) protein significantly affected the viral replication capacity and virulence of these two viruses. Taken together, our findings revealed the diverse genomic origins and virulence of the SIVs prevalent in Liaoning Province during 2014 to 2016, which highlights that continuous surveillance is essential to monitor the evolution of SIVs. We identified a naturally occurring amino acid mutation in the PB2 protein of H1N1 SIVs that impacts the viral replication and virulence in mice by altering the viral polymerase activity.IMPORTANCE The frequent reassortment among different influenza viruses in pigs adds complexity to the epidemiology of swine influenza. The diverse viral virulence phenotypes underline the need to investigate the possible genetic determinants for evaluating the pandemic potential to human public health. Here, we found that multiple genotypes of influenza viruses cocirculate in the swine population in Liaoning Province, China. Furthermore, we pinpointed a single amino acid at position 431 in the PB2 protein which plays a critical role in the virulence of H1N1 viruses in mice and found that the alteration of viral polymerase activities is the cause of the different virulence. Our study further indicated that the virulence of influenza virus is a polygenic trait, and the newly identified virulence-related residue in the PB2 provides important information for broadening knowledge on the genetic basis of viral virulence of influenza viruses.

Immune efficacy of an adenoviral vector-based swine influenza vaccine against antigenically distinct H1N1 strains in mice.

Avian-like H1N1 swine influenza viruses are prevalent in pigs and have occasionally crossed the species barrier and infected humans, which highlights the importance of preventing swine influenza. Human adenovirus serotype 5 (Ad5) has been tested in human influenza vaccine clinical trials and has exhibited a reliable safety profile. Here, we generated a replication-defective, recombinant adenovirus (designated as rAd5-avH1HA) expressing the hemagglutinin gene of an avian-like H1N1 virus (A/swine/Zhejiang/199/2013, ZJ/199/13). Using a BALB/c mouse model, we showed that a two-dose intramuscular administration of recombinant rAd5-avH1HA induced high levels of hemagglutination inhibition antibodies and prevented homologous and heterologous H1N1 virus-induced weight loss, as well as viral replication in the nasal turbinates and lungs of mice. Furthermore, a prime-boost immunization strategy trial with a recombinant plasmid (designated as pCAGGS-HA) followed by rAd5-avH1HA vaccine provided effective protection against homologous and heterologous H1N1 virus infection in mice. These results indicate that rAd5-avH1HA is an efficacious genetically engineered vaccine candidate against H1N1 swine influenza. Future studies should examine its immune efficacy in pigs.

Protective efficacy of an inactivated Eurasian avian-like H1N1 swine influenza vaccine against homologous H1N1 and heterologous H1N1 and H1N2 viruses in mice.

Eurasian avian-like H1N1 (EA H1N1) swine influenza viruses are prevalent in pigs in Europe and Asia, but occasionally cause human infection, which raises concern about their pandemic potential. Here, we produced a whole-virus inactivated vaccine with an EA H1N1 strain (A/swine/Guangxi/18/2011, SW/GX/18/11) and evaluated its efficacy against homologous H1N1 and heterologous H1N1 and H1N2 influenza viruses in mice. A strong humoral immune response, which we measured by hemagglutination inhibition (HI) and virus neutralization (VN), was induced in the vaccine-inoculated mice upon challenge. The inactivated SW/GX/18/11 vaccine provided complete protection against challenge with homologous SW/GX/18/11 virus in mice and provided effective protection against challenge with heterologous H1N1 and H1N2 viruses with distinctive genomic combinations. Our findings suggest that this EA H1N1 vaccine can provide protection against both homologous H1N1 and heterologous H1N1 or H1N2 virus infection. As such, it is an excellent vaccine candidate to prevent H1N1 swine influenza.

Immunogenicity and efficacy of a recombinant adenovirus expressing hemagglutinin from the H5N1 subtype of swine influenza virus in mice.

The H5N1 influenza viruses infect a range of avian species and have recently been isolated from humans and pigs. In this study we generated a replication-defective recombinant adenovirus (rAd-H5HA-EGFP) expressing the hemagglutinin (HA) gene of H5N1 A/Swine/Fujian/1/2001 (SW/FJ/1/01) and evaluated its immunogenicity and protective efficacy in BALB/c mice. The recombinant virus induced high levels of hemagglutination inhibition (HI) antibody at a median tissue culture infective dose of 10(8) or 10(7). Compared with mice in the control groups, the mice vaccinated with rAd-H5HA-EGFP did not show apparent weight loss after challenge with either the homologous SW/FJ/1/01 or the heterologous H5N1 A/Chicken/Hunan/77/2005 (CK/HuN/77/05). Replication of the challenge virus was partially or completely inhibited, and viruses were detected at significantly lower numbers in the organs of the vaccinated mice, all of which survived the challenge with CK/HuN/77/05, whereas most of the control mice did not. These results indicate that rAd-H5HA-EGFP can provide effective immune protection from highly pathogenic H5N1 viruses in mice and is therefore a promising new candidate vaccine against H5N1 influenza in animals.

Les virus de l'influenza de type H5N1 infectent une grande variété d'espèces aviaires et ont récemment été isolés des humains et des porcs. Dans la présente étude nous avons généré un adénovirus recombinant défectueux pour la réplication (rAd-H5HA-EGFP) exprimant le gène de l'hémagglutinine (HA) du virus H5N1 A/Porc/Fujian/1/2001 (SW/FJ/1/01) et évalué son immunogénicité et son activité protectrice chez des souris BALB/c. Le virus recombinant induisit des titres élevés d'anticorps inhibant l'hémagglutination (HI) à une dose médiane infectant les cultures cellulaires de 108 ou 107. Comparativement aux souris des groupes témoins, les souris vaccinées avec rAd-H5HA-EGFP n'ont pas montré de perte de poids apparente après une infection défi avec soit le virus homologue SW/FJ/1/01 ou le virus H5N1 hétérologue A/Poulet/Hunan/77/2005 (CK/HuN/77/05). La réplication du virus de l'infection défi fut partiellement ou complètement inhibée, et les virus furent détectés à une charge virale significativement inférieure dans les organes des souris vaccinées, qui ont toutes survécu à l'infection défi avec le virus CK/HuN/77/05, ce qui ne fut pas le cas de la majorité des souris témoins. Ces résultats indiquent que rAd-H5HA-EGFP peut fournir une protection efficace contre des virus H5N1 hautement pathogènes chez les souris et est ainsi un nouveau vaccin candidat prometteur contre l'influenza H5N1 chez les animaux.(Traduit par Docteur Serge Messier).