The volume of network traffic is increasing day by day. After the withdrawal of America in Iraq and the formation of national governments and the establishment of the network, the volume of network traffic in Iraq has increased significantly. Nevertheless, attacks and network intrusions are still considered a serious threat. There are various methods to solve these challenges. One of the most important methods is the use of intrusion detection algorithms (IDS). These methods have various advantages in intrusion detection; however, they still have problems. These methods usually have problems such as low accuracy, not being lightweight, and the use of complex and impractical algorithms. In recent years, hybrid methods have solved these problems, but not much attention has been paid to the type of algorithms used, as well as the optimization of hyperparameters. In this research, the goal is to create an accurate and lightweight hybrid intrusion detection method, consisting of deep learning and ANFIS network. The innovation of this research compared to the previous methods is the design of a cascade structure, based on which a suitable intrusion detection algorithm can be created accurately and in addition to achieving lightweight. Also, in the design of the deep network, we analyze the hyperparameters of the deep network to obtain the best hyperparameters and achieve more accuracy in intrusion detection. The simulation results on the NSL-KDD data set show that this method has 96.42% accuracy and good speed. This means that with the proposed structure, we can move towards an accurate and lightweight hybrid intrusion detection method, consisting of deep learning and ANFIS network. Therefore, the proposed method can be a practical method to detect intrusion in the national Internet platform of Iraq.
