Application of Adaptive Neural-Fuzzy Intelligent Models in Extension of Hybrid Force/Displacement Seismic Design method for EBF Regular Structures, subjected to Near-fault Earthquakes

Due to the significant advantages of the performance-based seismic design method, such as the possibility of determining the possible damage and financial and human losses of residents and neighbors of the structure, this method has been widely welcomed. However, since this method requires more sophisticated analysis than conventional force methods, sometimes the simple force method is preferred by some professional engineers. The main purpose of this article is to combine the two methods of force-based and performance-based and to develop a hybrid method in order to use the advantages of both methods.in this regard, frames with 3, 6, 9, 12, 15 and 20 story with 3 bays with a width of 5 meters have been considered. The length of the link beam is defined as another parameter affecting the response, 1, 1.75 and 2.50 meters. The studied models have been developed by designing the method of load and resistance factor design method, for 3 performance levels of immediate occupancy, life safety and collapse prevention, as well as the first occurrence of the plastic joint. The final models are analyzed under 20 pulse-type near-fault records using time history analysis. To generate the expected database, 12,960 time history analyzes were performed based on an incremental dynamic analysis platform. In this regard, a unique frame is continuously and repeatedly affected by a single accelerometer by multiplying the accelerometer by an SF coefficient. In each iteration, the maximum displacement in the frame is compared to the target range of ASCE41-13 code. The analysis operation is continued until the expected numbers are reached and then stopped. For each of the frames, 4 different acceptance levels are defined to consider different performance levels. Finally, using the genetic algorithm, the corresponding experimental relationships are presented to determine the behavior factor, local and global ductility. The proposed relationships are influenced by geometric characteristics such as the number of stories, the stiffness ratio of the columns, the slenderness of the braces, the length of the beam and the ductility levels. The first ambiguous issue that has been less mentioned in previous research is the use of near-fault field records in the development of a hybrid functional seismic design method. After generating 12960 data from an innovative time history analysis, two intelligent adaptive neural-fuzzy models have been used to calculate the coefficient of behavior and ductility of the structure. In order to create the best and most accurate model, Fuzzy C-Mean clustering (FCM) and Subtracting clustering methods have been used. Based on the results, the model created based on Subtracting clustering provides more accurate results than the other model. The results of hybrid seismic design in comparison with the force method and equivalent time history show the acceptable accuracy of the method introduced in the field of hypotheses. The obvious advantage of using a hybrid seismic design method compared to force methods is the possibility of selecting an expected performance level, which leads to design control and more accurate estimation of response values of quantities such as global ductility, local ductility, inter-story drift