Artigo em periódicoDestaque

Machine Learning-Based Intrusion Detection for Industrial Control Systems

Iury Valente de Bessa, Eddie Batista de Lima Filho, Lucas Carvalho Cordeiro, Renan Landau Paiva de Medeiros

2025 • IEEE Transactions on Dependable and Secure Computing • 8 citações

Resumo

This work proposes a machine learning-based intrusion detection system specifically designed for industrial control systems. We develop a hybrid approach combining supervised learning (Random Forest) with unsupervised learning (Autoencoder) to detect both known and zero-day attacks. The system is trained and tested using real network traffic data from an operational SCADA system in a water treatment facility. Experimental results show detection rates above 98% with low false positive rates below 2%, demonstrating the effectiveness of the approach for protecting critical infrastructure.

Autoria

Autores (2)

Iury Valente de Bessa

Co-líder atual do e-Controls

Lucas Carvalho Cordeiro

Docente da linha de pesquisa

projeto relacionado

Projeto Relacionado

Estabilidade e Controle de Microrredes CC

Projeto de pesquisa focado na análise de estabilidade e desenvolvimento de técnicas avançadas de controle para microrredes de corrente contínua com alta penetração de fontes renováveis.

Concluído

palavras-chave