Predicting the pseudo acceleration for near field condition and C type soil using ANFIS model
Maryam Hassaninia and Hassan Sharafia*
a: Assistant professor, School of Faculty Engineering, Razi University, Kermanshah, iran
In recent years, adaptive neuro-fuzzy inference systems have been applied successfully to different branches of geotechnical and earthquake engineering. In this study for predicting the pseudo acceleration, actual records of earthquake and neuro-fuzzy inference system have been applied. Most effective parameters of earthquake and site characteristics were considered as: magnitude of earthquake, closest distance from the recording site to ruptured area and PGA. In addition, period of time should be given to the network for predicting the pseudo acceleration. Different soil conditions provide different spectral shape; therefore a special type of soil is selected. If vs30 is between 375 m/s and 750 m/s, based on NEHRP code soil is categorize as C type. In this study, vs30 is between the mentioned range, and according to the NEHRP code the selected soil is called C type. ANFIS toolbox of software MATLAB and gbell membership function were used in this model. The training was performed with an improved hybrid method. Kind and number of membership functions are obtained by trial and error. Results of this study illustrate that neuro-fuzzy systems have a desirable capability of predicting pseudo acceleration.
Keywords: Pseudo Acceleration, Neuro-Fuzzy, Membership Functions, Earthquake, Period
Get the entire paper (pdf)
Go back to the TOC