A Digital Twin Framework for Cyber-Physical System Detection in Water Treatment Infrastructure
Faculty Mentor
Rami Haddad
Location
Russell Union 2052
Type of Research
On-going
Session Format
Oral Presentation
College
Allen E. Paulson College of Engineering & Computing
Department
Electrical and Computer Engineering
Abstract
Water treatment plants increasingly rely on networked sensing and supervisory control, yet many facilities operate with legacy infrastructure and limited cyber defenses, leaving critical processes vulnerable to cyber-physical attacks. While legal and regulatory frameworks recognize the societal impact of disrupting clean water access, recent incidents demonstrate that adversaries continue to target water utilities. This paper presents a digital twin (DT)–driven detection framework that couples a discrete-event simulation (DES) model of a water treatment process with a long short-term memory (LSTM) classifier for attack detection and categorization. A conventional treatment train is modeled to generate event-level process data under normal operation and under targeted attacks on chemical dosing pumps. Using five classes, the proposed DT+LSTM approach achieves over 92% accuracy, precision, recall, and F1-score in detecting and classifying simulated attacks. These results demonstrate the feasibility of DES-based DTs as cyber-physical systems for monitoring water treatment operations and motivate further development toward higher-fidelity plant models and broader attack coverage.
Program Description
.
Start Date
4-23-2026 10:30 AM
End Date
4-23-2026 10:45 AM
Recommended Citation
Chen, Jonas, "A Digital Twin Framework for Cyber-Physical System Detection in Water Treatment Infrastructure" (2026). GS4 Student Scholars Symposium. 110.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/110
A Digital Twin Framework for Cyber-Physical System Detection in Water Treatment Infrastructure
Russell Union 2052
Water treatment plants increasingly rely on networked sensing and supervisory control, yet many facilities operate with legacy infrastructure and limited cyber defenses, leaving critical processes vulnerable to cyber-physical attacks. While legal and regulatory frameworks recognize the societal impact of disrupting clean water access, recent incidents demonstrate that adversaries continue to target water utilities. This paper presents a digital twin (DT)–driven detection framework that couples a discrete-event simulation (DES) model of a water treatment process with a long short-term memory (LSTM) classifier for attack detection and categorization. A conventional treatment train is modeled to generate event-level process data under normal operation and under targeted attacks on chemical dosing pumps. Using five classes, the proposed DT+LSTM approach achieves over 92% accuracy, precision, recall, and F1-score in detecting and classifying simulated attacks. These results demonstrate the feasibility of DES-based DTs as cyber-physical systems for monitoring water treatment operations and motivate further development toward higher-fidelity plant models and broader attack coverage.
