Industrial Internet of Things (IIoT) has risen dramatically and blocked practically all requests due to the fast growth and expansion of brainy schemes and independent, energy-aware sensing plans. IIoT-based botnet assaults have increased, despite the limits of IIoT devices in terms of compute, storage, and connectivity. A scheme that can create profiles for malicious events via IoT networks is required to counteract this threat. To detect botnet attacks in IIoT networks, we offer a machine learning ensemble model that analyses IoT network behaviour characteristics and using ensemble learning to spot out-of-the-ordinary traffic caused by compromised IoT nodes. We also compare four distinct machine learning methods, including Generalized Additive Models (GAMs), Random Forests (RF), Multi-Layer Perceptron (MLP), and XGBoost, to define our IoT-based botnet detection strategy. We utilised the dataset, which includes logs of both benign IoT attack IIoT devices, to assess the quality of the projected model. The testing findings show that our optional model has a high detection accuracy of 96.42% against botnet assaults initiated from hacked IIoT devices. We also compare the proposed model’s ensemble findings to those of existing state-of-the-art methods, showing that they are superior.