Social network analysis is a fascinating field that helps us analyze social networks and understand their structure and dynamics. Link prediction is a crucial task in social network analysis, involving the forecasting of the likelihood that a new connection will form between two nodes within the network. Many machine-learning techniques, including supervised and unsupervised learning, have been used by researchers to increase the accuracy of link prediction. However, these methods are constrained by the inflexibility of manually designed features, which cannot adapt during the learning process. Additionally, creating these features can be both time-consuming and expensive. Recently, deep learning models have emerged in the field of machine learning and related areas, showcasing exceptional performance across a wide range of tasks. In this paper, we aim to develop a link prediction approach based on Graph Convolutional Network (GCN). The main idea is to design an innovative architecture tailored to the specific characteristics of the network data, optimizing hyperparameters, and employing advanced training techniques to enhance model performance.