Automated identification of steel weld defects,a convolutional neural network improved machine learning approach

This paper proposes a machine-learning-based methodology to automatically classify different types of steel weld defects,including lack of the fusion,porosity,slag inclusion,and the qualified(no defects)cases.This methodology solves the shortcomings of existing detection methods,such as expensive equipment,complicated operation and inability to detect internal defects.The study first collected percussed data from welded steel members with or without weld defects.Then,three methods,the Mel frequency cepstral coefficients,short-time Fourier transform(STFT),and continuous wavelet transform were implemented and compared to explore the most appropriate features for classification of weld statuses.Classic and convolutional neural network-enhanced algorithms were used to classify,the extracted features.Furthermore,experiments were designed and performed to validate the proposed method.Results showed that STFT achieved higher accuracies(up to 96.63%on average)in the weld status classification.The convolutional neural network-enhanced support vector machine(SVM)outperformed six other algorithms with an average accuracy of 95.8%.In addition,random forest and SVM were efficient approaches with a balanced trade-off between the accuracies and the computational efforts.

Analytical Solutions for Free Vibration and Buckling of Composite Beams Using a Higher Order Beam Theory

To satisfy the interfacial shear force continuity conditions, a new model is proposed for the two-layer composite beam with partial interaction by modifying Reddy＇s higher order beam theory. The governing differential equations for free vibration and buckling are formulated using the Hamilton＇s principle, the natural frequencies and axial forces are thus analytically obtained by Laplace transform technique. The analytical results are verified through the comparison with those of several other models common in use; and the presented model is found to be a finer one than the Reddy＇s. A parametric study is also performed to investigate the effects of geometry and material parameters.

Upconversion nanoparticles for super-resolution quantification of single small extracellular vesicles

Although small EVs(sEVs)have been used widely as biomarkers in disease diagnosis,their heterogeneity at single EV level has rarely been revealed.This is because high-resolution characterization of sEV presents a major challenge,as their sizes are below the optical diffraction limit.Here,we report that upconversion nanoparticles(UCNPs)can be used for super-resolution profiling the molecular heterogeneity of sEVs.We show that Er3+-doped UCNPs has better brightness and Tm3+-doped UCNPs resulting in better resolution beyond diffraction limit.Through an orthogonal experimental design,the specific targeting of UCNPs to the tumour epitope on single EV has been cross validated,resulting in the Pearson’s R-value of 0.83 for large EVs and~65%co-localization double-positive spots for sEVs.Furthermore,super-resolution nanoscopy can distinguish adjacent UCNPs on single sEV with a resolution of as high as 41.9 nm.When decreasing the size of UCNPs from 40 to 27 nm and 18 nm,we observed that the maximum UCNPs number on single sEV increased from 3 to 9 and 21,respectively.This work suggests the great potentials of UCNPs approach“digitally”quantify the surface antigens on single EVs,therefore providing a solution to monitor the EV heterogeneity changes along with the tumour progression progress.