Experimentation data of perspex glass sheet cutting, using CO2 laser, with missing values were modelled with semi-supervised artificial neural networks. Factorial design of experiment was selected for the verification...Experimentation data of perspex glass sheet cutting, using CO2 laser, with missing values were modelled with semi-supervised artificial neural networks. Factorial design of experiment was selected for the verification of orthogonal array based model prediction. It shows improvement in modelling of edge quality and kerf width by applying semi-supervised learning algorithm, based on novel error assessment on simulations. The results are expected to depict better prediction on average by utilizing the systematic randomized techniques to initialize the neural network weights and increase the number of initialization. Missing values handling is difficult with statistical tools and supervised learning techniques; on the other hand, semi-supervised learning generates better results with the smallest datasets even with missing values.展开更多
Due to the extreme complexity of mechanical response of soft surrounding rock(SR) around a tunnel under high geostatic stress conditions, the integration of physical and numerical modeling techniques was adopted. Base...Due to the extreme complexity of mechanical response of soft surrounding rock(SR) around a tunnel under high geostatic stress conditions, the integration of physical and numerical modeling techniques was adopted. Based on the similarity theory, new composite-similar material was developed, which showed good agreement with the similarity relation and successfully simulated physico-mechanical properties(PMP) of deep buried soft rock. And the 800 mm×800 mm×200 mm physical model(PM) was conducted, in which the endoscopic camera technique was adopted to track the entire process of failure of the model all the time. The experimental results indicate that the deformation of SR around a underground cavern possessed the characteristics of development by stages and in delay, and the initial damage of SR could induce rapid failure in the later stage, and the whole process could be divided into three stages, including the localized extension of crack(the horizontal load(HL) was in the range of 130 k N to 170 k N, the vertical load(VL) was in the range of 119 k N to 153.8 k N), rapid crack coalescence(the HL was in the range of 170 k N to 210 k N, the VL was in the range of 153.8 k N to 182.5 k N) and residual strength(the HL was greater than 210 k N, the VL was greater than 182.5 k N). Under the high stress conditions, the phenomenon of deformation localization in the SR became serious and different space positions show different deformation characteristics. In order to further explore the deformation localization and progressive failure phenomenon of soft SR around the deeply buried tunnel, applying the analysis software of FLAC3 D three-dimensional explicit finite-difference method, based on the composite strain-softening model of Mohr-Coulomb shear failure and tensile failure, the calculation method of large deformation was adopted. Then, the comparative analysis between the PM experiment and numerical simulation of the three centered arch tunnels was implemented and the relationship of deformation localization and progressive failure of SR around a tunnel under high stress conditions was discussed.展开更多
文摘Experimentation data of perspex glass sheet cutting, using CO2 laser, with missing values were modelled with semi-supervised artificial neural networks. Factorial design of experiment was selected for the verification of orthogonal array based model prediction. It shows improvement in modelling of edge quality and kerf width by applying semi-supervised learning algorithm, based on novel error assessment on simulations. The results are expected to depict better prediction on average by utilizing the systematic randomized techniques to initialize the neural network weights and increase the number of initialization. Missing values handling is difficult with statistical tools and supervised learning techniques; on the other hand, semi-supervised learning generates better results with the smallest datasets even with missing values.
基金Projects(51374257,50804060)supported by the National Natural Science Foundation of ChinaProject(NCET-09-0844)supported by the New Century Excellent Talent Foundation from MOE of China
文摘Due to the extreme complexity of mechanical response of soft surrounding rock(SR) around a tunnel under high geostatic stress conditions, the integration of physical and numerical modeling techniques was adopted. Based on the similarity theory, new composite-similar material was developed, which showed good agreement with the similarity relation and successfully simulated physico-mechanical properties(PMP) of deep buried soft rock. And the 800 mm×800 mm×200 mm physical model(PM) was conducted, in which the endoscopic camera technique was adopted to track the entire process of failure of the model all the time. The experimental results indicate that the deformation of SR around a underground cavern possessed the characteristics of development by stages and in delay, and the initial damage of SR could induce rapid failure in the later stage, and the whole process could be divided into three stages, including the localized extension of crack(the horizontal load(HL) was in the range of 130 k N to 170 k N, the vertical load(VL) was in the range of 119 k N to 153.8 k N), rapid crack coalescence(the HL was in the range of 170 k N to 210 k N, the VL was in the range of 153.8 k N to 182.5 k N) and residual strength(the HL was greater than 210 k N, the VL was greater than 182.5 k N). Under the high stress conditions, the phenomenon of deformation localization in the SR became serious and different space positions show different deformation characteristics. In order to further explore the deformation localization and progressive failure phenomenon of soft SR around the deeply buried tunnel, applying the analysis software of FLAC3 D three-dimensional explicit finite-difference method, based on the composite strain-softening model of Mohr-Coulomb shear failure and tensile failure, the calculation method of large deformation was adopted. Then, the comparative analysis between the PM experiment and numerical simulation of the three centered arch tunnels was implemented and the relationship of deformation localization and progressive failure of SR around a tunnel under high stress conditions was discussed.
