作者机构：
Institute of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan, 430070, Chin;Hunan City University, Yiyang, 413000, China
会议名称：
第三届中日岩土工程研讨会(Proceedings of the 3rd Sino-Japan Geotechnical Symposium)
摘要：
With the building of freeway, the problem of soft ground is becoming more and more protrusive and has been one of the key factors influencing project quality, project cycle and project cost. It has become a very important task to strengthen the research on the improvement of soft ground, which affects the development of traffic in China. In view of the incompleteness and uncertainty of information on soft ground improvement method,the decision model of the soft soil improvement methods based on multi-layer fuzzy is put forward. In this model, multi-layer structure model is set up, the weights of rule layer and index layer are given, the character index matrix of project layer is put forward and total ordering of the projects is decided. The actual examples results indicated that the decision model is reasonable, practical and convenient to use.
会议名称：
Nonintrusive Inspection, Structures Monitoring, and Smart Systems for Homeland Security
会议时间：
San Diego, CA, United States
会议论文集名称：
Nonintrusive Inspection, Structures Monitoring, and Smart Systems for Homeland Security
关键词：
acceleration;substructure;identification;neural networks;root mean square of prediction difference vector (RMSPDV);stiffness;damping coefficients
摘要：
A substructural identification methodology by the direct use of acceleration measurements with neural networks is proposed. The rationality of the substructural identification methodology employing a substructural acceleration-based emulator neural network (SAENN) and a substructural parametric evaluation neural network (SPENN) is explained. Based on the discrete time solution of the state space equation of the substructure, the theory basis for the construction of SAENN and SPENN is described. An evaluation index called root mean square of prediction difference vector (RMSPDV) corresponding to acceleration response is presented to evaluate the condition of object structure. The performance of the SAENN for acceleration forecasting and SPENN for parametric identification is examined by numerical simulations with a substructure of a 50-DOFs shear structure involving all stiffness and damping coefficient values unknown. Based on the trained SAENN and the PENN, the inter-storey stiffness and damping coefficients of the substructure are identified. Since the strategy does not require structural modes or frequencies extraction, it is computationally efficient, thus providing a possibly viable tool for structural identification and damage detection of large-scale infrastructures.
