In the design of the ordinary moment resisting frame to form plastic hinges in beams, the principle of weak-beam strong-column is used. Nevertheless, the plastic hinges take form in the beam and the vicinity of the column, the relocation of the plastic hinges is an adequate approach to increase the performance of the joint and also the structure. One of the methods to achieve this goal is strengthening the joints with FRP material. Joints in real structures deal with some limitation, such as sides beams which are connected to the joints at the floor level. These conditions make some problems to strengthen the joints. Therefore, the configuration of FRP application is considered in the form of L-shaped at the top and bottom beam, wrapping for beam and wrapping for the columns. Also the specification of moment-rotation of the joint strengthened with FRP is a necessity to determine the effects of strengthened joints on nonlinear modeling of the frame. In this study, two types of external RC joints are examined. The first category of joints that based on ACI-318 was designed for the moderate ductility. The second category is similar to the first joints, but to improve the behavior joints CFRP sheets was used. After the numerical model of joint was confirmed by experimental model, a total of 216 connections under monotonic load is simulated. Since the numeric modeling of the joint is a very time consuming and expensive procedure, therefore training of a neural network to predict the moment-rotation of the strengthened joint with FRP is another necessity. In this research, two neural networks are trained to predict the moment-rotation curves of the original joints and the strengthened joint using the moment-rotation curves of these joints. The results show moment-rotation curves obtained from the numerical modeling and moment-rotation curves obtained from the neural networks have acceptable accuracy.
