Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/7558
Title: COMPARISON OF SITE SPECIFIC PGA USING NEURAL NETWORKS AND REGRESSION MODELS
Authors: Venugopal, S.
Keywords: EARTHQUAKE ENGINEERING;SITE SPECIFIC PGA;NEURAL NETWORKS;REGRESSION MODELS
Issue Date: 2010
Abstract: Knowing the characteristics of ground motions in a specified region is vital for the design of engineering structures. Peak ground acceleration (PGA) is commonly used to define the ground motions. One method for estimating these loading conditions is through equations based on strong ground motion recorded during previous earthquakes. Although the equations are often referred to as attenuation relationships, attenuation relations or attenuation equations, they predict more than how ground motion varies with distance. As result of this investigation very little agreement has been reached in the past 30 years of ground motion estimation relation studies since each formula had been derived based on the available data which varied greatly with geographical regions.Artificial neural networks (ANNs), however, are not defined as a specific equation form. They can infer solutions to problems having nonlinear and complex interaction among the variables and find functional relationship between the input and output of dataset. In this study, an attempt has been made to develop efficient neural network based models tt predict peak ground acceleration for different soil categories based on FEMA 356, 2000. Which is of significance in earthquake engineering applications, and the comparison of PGA. with (NGA model ) Boore and Atkinson, (2008), attenuation model developed by boore et. (1997), Sharma (1998), Fukushima et al, (1990) with the developed neural network models. In this study, sets of Multilayer Perceptron neural network model are trained to predict the Peak Ground Acceleration (PGA), using Japanese earthquake records and site characteristics. The database used in the study is taken from Kyoshin Net (K-NET) database of Japan. Neuro Intelligence software has been used to model ANN and the standard back-propagation supervised training scheme is used to train all networks. ANN has been used to solve the problem of predicting Peak Ground Acceleration using records of Japanese earthquakes of magnitude more than 4.0 and hypocentral distance less than 200 km. In this study, 37,211 horizontal components of time histories have been used, and these time histories were divided into five categories i.e. (5 data sets) based on the average shear wave velocity (vs) as per FEMA 356, 2000. Also all time histories having PGA less than 10 cm/sect were removed from the data set. Basic information such as magnitude hypocentral distance, shear wave velocity, and fault type are used as input variables to train the neural network. Five neural network models are developed for predicting PGA for the five datasets (i,e Class A, B, C, D and E) and the adequacy of the models was evaluated on basis of Network error, correlation coffecient and percentage error. ANN model could predict Accurate values (percentage error less than 3 %) for as high as 48 %.Estimation of PGA using the attenuation relationships developed by Boore and Atkinson (2008), Sharma (1998), Boore et al. (1997), Fukushima & Tanaka (1990) has been done. Then correlation coffecient of Attenuation models are compared with the ANN models developed which is as high as 0.98 and low is 0.92 for ANN, 0.61 high and 0.31 is low for attenuation models.PGA is estimated for distance ranging from (10-200km) and plotted for the magnitudes 7, 6 and 5 and to the Class A, B, C, D, and E (5 Data sets) using ANN models and Attenuation relationships. Then the comparative study has been made. From the results presented and discussed in this study, it can be concluded that the ANN is a valuable tool for prediction of strong ground motion parameters such as Peak Ground Acceleration given the earthquake source, path and local soil conditions.
URI: http://hdl.handle.net/123456789/7558
Other Identifiers: M.Tech
Research Supervisor/ Guide: Sharma, M. L.
Kumar, Ashok
metadata.dc.type: M.Tech Dessertation
Appears in Collections:MASTERS' THESES (Earthquake Engg)

Files in This Item:
File Description SizeFormat 
EQD G20287.pdf8.41 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.