A Siamese hybrid neural network framework for few-shot fault diagnosis of fixed-wing unmanned aerial vehicles

Li Shaobo,Li Chuanjiang,Zhang Ansi,Yang Lei,Zio Enrico,Pecht Michael,Gryllias Konstantinos 한국CDE학회 2022 Journal of computational design and engineering Vol.9 No.4

As fixed-wing unmanned aerial vehicles (FW-UAVs) are used for diverse civil and scientific missions, failure incidents are on the rise. Recent rapid developments in deep learning (DL) techniques offer advanced solutions for fault diagnosis of unmanned aerial vehicles. However, most existing DL-based diagnostic models only perform well when trained on massive amounts of labeled data, which are challenging to collect due to the complexity of the FW-UAVs systems and service environments. To address these issues, this paper presents a novel framework, Siamese hybrid neural network (SHNN), to achieve few-shot fault diagnosis of FW-UAVs in an intelligent manner. “State map” strategy is firstly proposed to transform raw flight data into similar and dissimilar sample pairs as input. The proposed SHNN framework consists of two identical networks that share weights with each other, and each subnetwork is designed with a hybrid one-dimensional conventional neural network and long short-term memory model as feature encoder, whose generated feature embedding is used to measure the similarity of input pairs via a distance function in the metric space. In comprehensive experiments on a real flight dataset of an FW-UAV, the SHNN framework achieves competitive results compared to other models, demonstrating its effectiveness in both binary and multi-class few-shot fault diagnosis.

Characterization of genes coding for galacturonosyltransferaselike (GATL) proteins in rice

Shaobo Li,Jinlong Liu,Mansi Luo,Xin Yan,Chao Yu 한국유전학회 2016 Genes & Genomics Vol.38 No.10

In the present study, seven galacturonosyltransferase- like (GATL) genes (OsGATLs) in rice (Oryza sativa L.) were genome-widely identified and the chromosomal locations and the gene structures of which were characterized. Under normal condition, OsGATL2 and OsGATL3 are highly expressed in root, while OsGATL4 is highly expressed in stem and leaf. Many cis-elements related to stress response and plant hormone were found in the promoter sequence of each OsGATL. The expression patterns of these OsGATL genes under treatment with abscisic acid (ABA), drought and low temperature were assessed by qRT-PCR. The expression levels of most OsGATLs significantly increased following the treatments with drought or low temperature. In addition, physicochemical properties of OsGATLs and phylogenetic analysis with GATL from rice and several other species were performed. 3D structures and protein–protein interaction (PPI) network of OsGATLs were further predicted by Swiss-model and STRING 9.0 database, respectively. The identification and bioinformatic analysis of GATL family in rice could provide reference data for further study on their biological functions, especially in the responsiveness to hormones and stress signaling.

Mechanism of MnS Precipitation on Al2O3–SiO2 Inclusions in Non‑oriented Silicon Steel

Fangjie Li,Huigai Li,Di Huang,Shaobo Zheng,Jinglin You 대한금속·재료학회 2018 METALS AND MATERIALS International Vol.24 No.6

This study investigates the mechanism of MnS precipitation on Al2O3–SiO2 inclusions during the solidification of nonorientedsilicon steel, especially the influence of the phase structures and sizes of the oxides on the MnS precipitation, byscanning electron microscopy and transmission electron microscopy coupled with energy dispersive spectrometry. Theinvestigation results show that MnS tends to nucleate on submicron-sized Al2O3–SiO2 inclusions formed by interdendriticsegregation and that it covers the oxides completely. In addition, MnS can precipitate on micron-sized oxides and its precipitationbehavior is governed by the phase structure of the oxides. The MnS embryo formed in a MnO-containing oxidecan act as a substrate for MnS precipitation, thus permitting further growth via diffusion of solute atoms from the matrix. MnS also precipitates in a MnO-free oxide by the heterogeneous nucleation mechanism. Furthermore, MnS is less proneto precipitation in the Al2O3-rich regions of the Al2O3–SiO2 inclusions; this can be explained by the high lattice disregistrybetween MnS and Al2O3.

Ensemble of Classifiers Constructed on Class-Oriented Attribute Reduction

Min Li,Shaobo Deng,Lei Wang 한국정보처리학회 2020 Journal of information processing systems Vol.16 No.2

Many heuristic attribute reduction algorithms have been proposed to find a single reduct that functions as theentire set of original attributes without loss of classification capability; however, the proposed reducts are notalways perfect for these multiclass datasets. In this study, based on a probabilistic rough set model, we proposethe classoriented attribute reduction (COAR) algorithm, which separately finds a reduct for each target class. Thus, there is a strong dependence between a reduct and its target class. Consequently, we propose a type ofensemble constructed on a group of classifiers based on classoriented reducts with a customized weightedmajority voting strategy. We evaluated the performance of our proposed algorithm based on five real multiclassdatasets. Experimental results confirm the superiority of the proposed method in terms of four generalevaluation metrics.

Experimental Study on the Damage of Steel Tubular Structural Components by Near-Field Detonations

Wanyue Wang,Shaobo Geng,Hua Wang,Wenqiang Li,Yaling Liu,Jianying Xue,Tingbian Zhan,Ying Gao 대한토목학회 2021 KSCE Journal of Civil Engineering Vol.25 No.2

This paper describes three blast-loading trials on three kinds of steel tubular structural components at the same explosive charge and standoff distance. Specimen one is a hollowsteel tube (HST), another is a HST wrapped with glass fibre-reinforced plastic (GFRP) with epoxy resin and the third is a HST infilled with concrete. The main objective of the trials is to investigate the effect of near-field detonations on circular steel tubular components. The experimental data, such as the overpressure time history, front local deformation, rear residual deflection and strain time history, are all recorded and collected. Analysis of the trial results shows that the experimental peak overpressure values of shot 1-3 are all larger than that of numerical simulation. The failure of these three specimens mainly experiences local damage, at the same time, the HST and HST with GFRP exhibit obvious global deformation. With the initiation point at one end of the cylinder explosive, both the maximum depth deformation on the front surface and the maximum residual deflection on the rear surface are all located on the side of the another end of the cylinder explosive. The blast-resistant approach by covering 10-layer GFRP on the surface of the HST can reduce the local damage, the global deformationand the dynamic strain value to certain extent, while the approach by infilling the steel tube with concrete can greatly decrease the local damage and the dynamic strain value and without any global deformation.

The “living” Feature of the ATRP Macroinitiators in Different Catalytic Systems

Yazhen Wang,Li Liu,Shaobo Dong,Xilai Zhou,Chenglong Wang,Zhen Shi 대한금속·재료학회 2021 ELECTRONIC MATERIALS LETTERS Vol.17 No.2

Atom transfer radical polymerization (ATRP) has achieved widespread use in living polymerization. However, until now there hasbeen little report that macroinitiators initiate polymerization in different catalytic systems. The preparation of bromine-terminatedpolymethyl methacrylate (PMMA-Br) and chlorine-terminated PMMA (PMMA-Cl) were carried out via reverse atom transferradical polymerization (RATRP). The PMMA with halogen termination and narrow polydispersity (Mn = 12,000–15,000 g/mol,Mw/Mn = 1.1–1.2) were used as macroinitiators. The block copolymer of polymethyl methacrylate and polyacrylonitrile (PMMAb-PAN) was prepared in different catalytic systems through normal ATRP. The analyses of the 1H NMR showed that the PMMAprepared by RATRP were end-functionalized by halogen atoms, demonstrated the activities of the PMMA macroinitiators. Themolecular weight and polydispersity index (PDI) of the polymers were analyzed using gel permeation chromatography (GPC). The results indicated that the block polymers that the molecular weight of the block copolymer after chain extension has increasedsignificantly and the molecular weight distribution is narrow (Mn = 17,000–25,000 g/mol, Mw/Mn = 1.1–1.3). The kinetics ofthese polymerization processes were studied as a function of monomers to the macroinitiator molar ratio. It was found that thepolymerizations in different catalytic systems coincidence first-order kinetics with respect to monomers.

Ultrathin Biocompatible Electrospun Fiber Films for Self-Powered Human Motion Sensor

Gengrui Zhao,Shaobo Gong,Honggang Wang,Junfang Ren,Na Wang,Yawen Yang,Gui Gao,Shengsheng Chen,Linlin Li 한국정밀공학회 2021 International Journal of Precision Engineering and Vol.8 No.3

In recent years, triboelectric nanogenerator (TENG) has been proven to be an effective and simple solution for self-powered wearable sensing, and a wide variety of materials and methods have been applied to TENG based sensors. In this study, an ultrathin fiber film with nanoporous brain-like structure (BUF) was obtained by cold-pressing the electrospun biocompatible fiber mats, and a wearable TENG sensor based on the BUF films that could sensitively detect human motion was demonstrated. This simply designed self-powered sensor was composed of Kapton as the shell and ultrathin, biocompatible poly L-lactic acid and ethyl cellulose (EC) fiber films as the triboelectric layers, prepared by convenient electrospinning method and cold-pressing. Compared to as-spun fiber mats, cold-pressed fiber membranes not only acquire overall compact structure but also remained porous surface morphology, which improved the triboelectric outputs and operational stability of TENG. Three TENGs were fabricated to evaluate the effects of different triboelectric layers on triboelectric outputs, and the BUF-TENG had the maximum triboelectric outputs (19 V, 630 nA at 3.5 Hz) and longer life than as-spun fiber mats based TENG. The BUF-TENG had linear relationship between triboelectric outputs and frequency, and sensitivities of 32.4, 0.94 and 0.22 V Hz−1 in the range of 0.1–0.35, 0.35–2 and 2–3.5 Hz, respectively. Moreover, a self-powered sensor based on the BUF-TENG was proposed to sensitively detect human walking state and limbs motion. With convenient preparation process and biocompatible materials, the BUF-TENG sensor could be a potential self-powered wearable and portable sensor in future health care and monitoring field.

Discovery of Subject of Science and Technology Policy based on LDA Model

Jianmin Wang,Shaobo Li,Tongrang Fan,Hui Wang 보안공학연구지원센터 2016 International Journal of Database Theory and Appli Vol.9 No.12

With the continuous increase of science and technology policy, how to find valuable information from the massive scientific and technological policies has becoming an urgent problem to be solved. Thus, this paper proposes a subject discovery method for large scale science and technology policy set. Based on LDA subject model for science and technology policy document subject modeling, this approach extracts time and geographical labels of science and technology policy, computes intensity of subjects under different time and geographical conditions, and obtains important subjects and analysis of theme change trend of the intensity of subjects under condition constraints. Experimental results demonstrate that this method can excavate and analyze the subjects form large scale science and technology policies quickly and effectively.

Influence of Cu2S Precipitates Dissolution on Ferrite Grain Growth During Heat Treatment in the Non-Oriented Electrical Steel Sheet

Yuan Wu,Fangjie Li,Ting Wang,Dan Zhao,Hefei Huang,Huigai Li,Shaobo Zheng 대한금속·재료학회 2017 METALS AND MATERIALS International Vol.23 No.3

The factor to deduce grain growth of non-oriented electrical steel sheet during heat treatment was researched in thispaper. Scanning electron microscope equipped with electron backscatter diffraction (EBSD), X-ray diffractionand transmission electron microscope were used to characterize the microstructure, dislocation density and precipitate,respectively. The EBSD results indicated that the grain size increased from 18.7 μm to 56 μm after heattreatment. Meanwhile, the characterization of grain size distribution, dislocation density and precipitates revealedthat the dissolution of Cu2S precipitates, which act as inhibitor, may lead to the grain growth. In addition, the pinningforce of grain boundary and the driving force of grain growth were calculated. Based on experiments resultsand theoretical calculations, the Cu2S precipitates with diameter of less than 39 nm and volume fraction of 1.74 ×10-4would be sufficient to inhibit the ferrite grain growth. Heat treatment provides an efficient way to prompt thegrain growth.

Inhibition of Tumor Growth by Recombinant Adenovirus Containing Human Lactoferrin through Inducing Tumor Cell Apoptosis in Mice Bearing EMT6 Breast Cancer

Jianjie Wang,Qingwang Li,Yetao Ou,Zengsheng Han,Kun Li,Peijun Wang,Shaobo Zhou 대한약학회 2011 Archives of Pharmacal Research Vol.34 No.6

Human lactoferrin (hLTF), an 80-kDa iron-binding glycoprotein, has antitumor activity. In this study, a recombinant adenovirus containing the human lactoferrin cDNA (ad-rhLTF) was constructed and its effect on tumor growth was investigated in mice bearing EMT6 breast cancer. Ad-rhLTF was injected seven times within 14 days into the tumor site at two concentrations (10^8 and 5 × 10^8 pfu/mL) in mice bearing EMT6 breast cancer. Injected ad-rhLTF had considerable cytotoxicity on mice breast cancer, and significantly reducing the weight of tumor produced and increasing the tumor inhibition rate up to 52.64%. The presence of apoptotic cells was confirmed using TUNEL staining and flow cytometry assays. At the same time, RTPCR and Western blot analyses demonstrated that ad-rhLTF also decreased expression of Bcl-2 and increased Bax and caspase 3 expressions. Therefore, we conclude that ad-rhLTF inhibits tumor growth by inducing tumor cell apoptosis in mice with breast cancer by triggering the mitochondrial-dependent pathway and activation of caspase 3. The results indicate that adrhLTF might be a promising drug for breast cancer gene therapy.