Cyberattacks are becoming increasingly complex and frequent, necessitating the importance using intrusion detection systems (IDS) to safeguard network perimeter. The advancement of Artificial Intelligence (AI) techniques has resulted in potential improvement in the effectiveness of IDS for detecting attacks. IoT is a significant contributor to increased data traffic and attack sources. It is necessary to extract significant features from the network traffic to construct the robust IDS model. So, this work aims to detect the IoT related attacks effectively on large volumes of network traffic using Improved Grey Wolf optimizer (IGWO) with a Vector Convolutional Network (VCN). The IGWO is proposed to select the important features from the large number of features to build the robust IDS using VCN. The proposed approach is experimented on two datasets: Bot-IoT and NSL-KDD. Bot-IoT which contains network traffic data generated by IoT devices, while NSL-KDD is a standard benchmark dataset for building the IDS. The proposed method effectively handles large volumes of network traffic data and shows better performance on both datasets.