In many different applications, wireless sensor network (WSN) security enhancement is crucial. The progress in localizing routing attacks is a critical focus of security research.  The network performance is compromised by introducing rogue nodes into wireless sensor networks, resulting in various routing attacks.  This research introduces a method called hybrid self-organizing neural networks (Hyb_SONN) for detecting and localizing numerous assaults.  The proposed approach consists of two main components: localization techniques and a neural network for the detection and localization of WSN DoS attacks. Using the Hyb_SONN method, the dataset is divided into training and testing sets to detect 10 different types of attacks, including denial-of-service (DoS) attacks. The purpose of the simulations is to assess the security-related effectiveness of the proposed technique for locating and detecting malicious nodes.  With little localization error, this method provides an accurate assessment of the unknown nodes’ position.  The proposed system can effectively detect and precisely locate malicious attacks in hierarchically distributed, scalable WSNs. The results indicate that the proposed approach has a high level of accuracy, with a rate of 99.8%. Additionally, the precision is measured at 96.4%, recall at 94.5%, PDR at 98.5%, energy consumption at 43.2%, and localization error at 13.2%.