Fatigue damage is one of the most common failure modes of large-scale engineering equipment, especially the full-face tunnel boring machine with characteristics of a thick plate structure bearing strong impact load. It is difficult to predict the location and propagation life of crack of cutterhead under strong impact load. Unseasonal maintenance of equipment caused by inaccurate prediction of life cycle of cutterhead seriously affects the construction efficiency of the equipment and the life safety of the operators. Determining the crack location of tunnel boring machine cutterhead structure under strong impact load and predicting the crack propagation life are difficult scientific problems. To solve them, first, the location of the stress concentration of the cutterhead is determined by using finite element analysis method of statics. Second, prediction model for crack propagation life of tunnel boring machine cutterhead characteristic substructure based on time integration is built. And the test of crack growth of cutterhead characteristic substructure is performed. The feasibility and accuracy of the prediction model are verified by contrasting crack prediction models and the results of the test. Finally, the life prediction of tunnel boring machine cutterhead of water diversion project in Northwest Liaoning Province is carried out by using crack propagation model based on time integration. Results show that the maximum error of theoretical prediction and experimental results of crack propagation is 16%. So the feasibility of crack propagation model based on time integration in predicting the crack growth of cutterhead is verified. It is predicted that the tunnel boring machine cutterhead panel can work normally for 5.9 km under the condition of ultimate load. Building the crack propagation model considering the influence of plate thickness and strong impact load has important research value for improving the working efficiency of engineering equipment, prolonging service time, and improving the working safety.
Through-focus scanning optical microscopy (TSOM) is one of the recommended measurement methods in semiconductor manufacturing industry in recent years because of its rapid and nondestructive properties. As a computational imaging method, TSOM takes full advantage of the information from defocused images rather than only concentrating on focused images. In order to improve the accuracy of TSOM in nanoscale dimensional measurement, this paper proposes a two-input deep-learning TSOM method based on Convolutional Neural Network (CNN). The TSOM image and the focused image are taken as the two inputs of the network. The TSOM image is processed by three columns convolutional channels and the focused image is processed by a single convolution channel for feature extraction. Then, the features extracted from the two kinds of images are merged and mapped to the measuring parameters for output. Our method makes effective use of the image information collected by TSOM system, for which the measurement process is fast and convenient with high accuracy. The MSE of the method can reach 5.18 nm2 in the measurement of gold lines with a linewidth range of 247–1010 nm and the measuring accuracy is much higher than other deep-learning TSOM methods.
Affected by complex boundary conditions and redundant structural assembly, it is difficult to collect stress and strain information on key structures of complex mechanical equipment, resulting in difficult life monitoring and large calculation errors. The purpose of this study is to provide reliable strain data for accurate life prediction and solve the problem of obtaining critical structural strain under uncertain loads and other complex boundary conditions. In this paper, a neural network training dataset is established based on finite element analysis data and experimental data, and the neural network architecture and parameter optimization are carried out for each data set. Finally, the strain reconstruction model based on a neural network is established, which solves the two problems of accurate equivalence from structure to strain and equivalence from strain to load. Feature experimental samples were designed to verify the universality of the research method. The model is experimentally verified with an average error of less than 5%. It realizes the global strain reconstruction from the local measurement point to the overall structure, which can be used for real-time and full-cycle life monitoring of the structure;
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.