An optimized anomaly detection framework in industrial control systems through grey wolf optimizer and autoencoder integration.

Ensuring reliable Internet connectivity in Industrial Control Systems is critical for real-time monitoring and anomaly detection. Existing methods, however, struggle with high computational complexity, limited applicability to specific datasets, and elevated false-positive rates. This paper presents a novel collaborative data processing framework that enhances anomaly detection in ICS by integrating the Grey Wolf Optimizer with Autoencoders. The proposed approach optimizes GWO by improving prey selection, encircling mechanisms, and initial population generation, while enhancing AE dropout functionality for improved model generalization. The method operates in two stages: (1) Optimizing GWO for feature selection to identify relevant features and reduce feature errors, and (2) Utilizing AE for efficient anomaly detection. Experimental validation on the SWaT and WADI benchmark datasets demonstrates the superior performance of the proposed model, achieving significant improvements in accuracy, precision, recall, and F1-score over existing state-of-the-art approaches. These results highlight the potential of the proposed approach in addressing the limitations of current anomaly detection systems in ICS.