Artificial Intelligence based Security Assessment Platform for Industrial Control Systems

Industrial Control Systems (ICS) serve as critical infrastructure sectors such as power generation, energy, and manufacturing by enabling real-time data acquisition, equipment monitoring, and automated control to ensure efficient industrial operations. As ICS evolves toward intelligence and openness—incorporating technologies like digital twins and AI and integrates with cross-regional collaborative industrial cloud platforms, it faces escalating cybersecurity challenges. The expanding attack surface, device distribution complexity, and stealthy nature of threats have led to a surge in risks such as data breaches, equipment downtime, and system paralysis. Notably, the number of attacks targeting ICS has grown exponentially in recent years. To address these challenges, this paper proposes a security evaluation platform tailored for intelligent industrial applications. The platform enables risk visualization management through log analysis of industrial systems, thereby enhancing the holistic security posture of ICS. Experimental results demonstrate that, compared to existing methods, our platform has high recognition rates at anomaly ratios of $15 \%$ and $20 \%$, indicating feasibility. Especially in 500 logs, the recognition rate is $100 \%$ for $20 \%$ anomaly ratio, 1000 logs, and $15 \%$ anomaly ratio. In addition, in 2500 logs, the $15 \%$ anomaly ratio is as high as $14.9 \%$, and in the case of $20 \%$ anomaly ratio, it is as high as $19.3 \%$.