The artificial neural network models have drawn engineer’s interest in recent years. The engineers need to study the kind and the feature of artificial neural network, which include the principle of fundamental and the frame of development, and to find the application of engineering. The paper introduces the artificial neural network which includes the operation manner and the process of evolvement and often uses these models and software, and presents the advantage and disadvantage between the artificial neural network and the traditional techniques, and review many results of study that have been used for geotechnical engineering. The effectiveness and applications of the artificial neural network are also presented in this paper.
Increasing use of pile foundations in Taiwan, an efficient method to evaluate pile lateral capacity has been drawn attention to geotechnical engineers. In general, we often need in-situ pile lateral load test to obtain p-y curves for capacity evaluation. This paper is to use the neural network method, via available in situ soil data, pile geometry and material properties, to estimate the load-displacement curve at pile head for laterally loaded piles.
In this study, numerous tunnel design case histories in Taiwan during the past 20 years were collected and utilized in order to establish a tunnel support design database. The database is further being analyzed by algorithms of artificial intelligence technologies such as case based reasoning (CBR) as well as artificial neural network (ANN). The results show that, with carefully selected tunnel case histories and adequate data mining method, rational designs of tunnel support works can be obtained by finding the most similar case or cases in the database. It this paper, only the database based on past experiences in the category of those tunnel excavated in the jointed rock mass have been studied.
Due to the complexity and uncertainty of factors affecting the underground construction, utilization of previous experiences as guidance in solving new problem is the major trend of engineering design for such projects. The rapidly developing artificial intelligence technologies, including expert system, case-based reasoning and artificial neural network, can provide an effective way of solving new problem by using experts’ experiences or engineering case histories. Based on the preliminary results of applying the artificial neural network to rock tunnel support design and prediction of diaphragm wall deflection in braced excavation, it appears that the artificial neural network can be a viable method in providing design recommendation or predicting engineering performance by using the previous case histories or monitoring data.
Predicting and analyzing engineering parameters from in-situ tests is an important and challenging task for geotechnical engineers. In this study, an artificial neural network approach is proposed to predict these design parameters. In this paper, a brief introduction of back-propagation networks is provided and then a network for predicting relative density Dr from CPT measurements is established. Discussions on the established network are presented, along with comparisons of the results by this network with existing methods. Meanwhile, the application of artificial neural network approach on the spatial variability characteristic of in-situ tests is also discussed in this study. Some details of the development of various network models for analyzing CPT measurements and the comparisons of the network predictions are presented.
In this paper, the diaphragm wall deflections of case histories and the construction stages of the concerned project were collected as the input database of the artificial neural network (ANN) with the error back-propagation (BP) model for the learning algorithm, and thus the wall deflections can be predicted for the next construction stage. Furthermore, the stress of the wall and the settlement of the ground evaluated by theoretical calculation or empirical formulae can be used as the basis for safety assessment. Case histories were adopted to verify the agreement of wall deflection between predicted and measured data.
Bared on the theory of plate tectonics, neural network, geologic hazards analysis and the remote sensing and monitoring of geotechnical techniques, it also finishes about the map of seismotectonics in Taiwan area, seismic risk map form 2001 to 2010 in Taiwan areas, the yearly tendency analysis of seismicity and the harmful analysis of seismic at Chiayi city.