The connectivity and integration of commodities have transformed many industries through the Internet of Things, enhancing both efficiency and functionality. However, this interconnection has also introduced critical security challenges, necessitating a robust Intrusion Detection System. This paper presents a novel, enhanced method for Intrusion Detection Systems in Internet of Things environments, utilizing advanced machine learning techniques and optimization algorithms. The proposed method integrates the Marine Predator Optimizer and Grey Wolf Optimizer for efficient feature selection, improving the detection and classification of anomalies. Two datasets, NFC-SECICIDS2018v2 and BotIoT2018, were used to evaluate the performance and effectiveness of the proposed Intrusion Detection System. The results demonstrate that the proposed method achieves an accuracy of 97.59% on the NFC-SECICIDS2018v2 dataset and 99.97% on the BotIoT2018 dataset. Additionally, the method shows a perfect sensitivity of 100% on the BotIoT2018 dataset and a sensitivity of 99.58% on the NFC-SECICIDS2018v2 dataset, significantly reducing the false alarm rate. The findings underscore the effectiveness of combining the Marine Predator Optimizer and Grey Wolf Optimizer to enhance the performance of Intrusion Detection Systems, offering a promising solution for securing the Internet of Things networks against evolving cyber threats ​