Learning Models of Cyber-Physical Systems with Discrete and Continuous Behaviour for Digital Twin Synthesis

Project Overview & Abstract

This project presents an interactive web-based platform for synthesizing and simulating a behavioral model for a Cyber-Physical System (CPS), a core component of a Digital Twin. The application provides an end-to-end workflow for a user to upload sensor and operational data, compare the performance of various machine learning and deep learning algorithms (Random Forest, SVM, PyTorch Neural Network), and deploy the best model in a live, interactive simulation. The system models both continuous (e.g., temperature, pressure) and discrete (e.g., operating mode) behaviors to accurately predict the overall state of the CPS (e.g., 'Stable', 'Warning', 'Failure'). This serves as a practical implementation of the concepts discussed in the base paper, moving from theoretical anomaly detection to a hands-on tool for model creation and real-time visualization.