Russian Scientists Develop Neural Network to Protect Industrial Systems
A self-learning AI model can detect cyberattacks and technical failures before they disrupt production.
A team of researchers from South Ural State University has created an AI-based security system designed to protect industrial facilities from failures and cyberattacks. The model uses a Kohonen neural network — an unsupervised learning algorithm that maps and recognizes patterns in complex data.
The system first studies the normal operational behavior of an industrial site and builds a baseline model. It then continuously monitors incoming signals, comparing them to this digital “fingerprint.” When it detects deviations, the neural network instantly flags potential threats.
94% Accuracy in Real-World Tests
The results, published in the proceedings of the International Conference on Industrial Engineering, Applications and Manufacturing, show that the model correctly classified 94% of test data, successfully identifying actions resembling cyberattacks.
The research team plans to increase the system’s accuracy and expand its ability to recognize a broader range of attack and malfunction scenarios. As industrial cybersecurity becomes one of the global IT sector’s most pressing challenges, the Russian project offers a modern, adaptive solution that could help safeguard critical infrastructure worldwide.
