Edge computing is a decentralized computing paradigm that processes data closer to its source, reducing latency and bandwidth usage while enabling real-time decision-making in applications such as IoT, autonomous systems, and industrial control. However, this distributed nature introduces significant security challenges, as edge devices are often resource-constrained and vulnerable to various cyberattacks. Traditional centralized intrusion detection systems (IDS) are ineffective in such environments due to scalability issues and high communication overhead. To address these limitations, this paper proposes an Adaptive Aggregated Federated Learning (AAFL) framework for intrusion detection in edge computing environments. The AAFL framework enables local model training on edge devices, ensuring data privacy and minimizing bandwidth usage by sharing only aggregated model updates with a central server. The framework incorporates adaptive aggregation, where the contributions of each device are weighted based on its model’s accuracy and training stability, ensuring that reliable devices have a greater influence on the global model. The proposed system enhances intrusion detection accuracy and system robustness by overcoming challenges such as device heterogeneity and unreliable updates. Experimental results show that the AAFL framework outperforms traditional federated learning models, providing a scalable and efficient solution for real-time intrusion detection in edge and IoT networks.