Edge AI Prognostics
AeroTwin: B787 Prognostics
Three-layer edge architecture running a zero-leakage CNN-BiLSTM with an offline Ollama 3.2 LLM mapping neural outputs to ATA 32 maintenance directives.
> Live high-fidelity video demonstrations of aerospace digital twins, cybersecurity architectures, and AI fluid simulations.
Three-layer edge architecture running a zero-leakage CNN-BiLSTM with an offline Ollama 3.2 LLM mapping neural outputs to ATA 32 maintenance directives.
Live demonstration of the blockchain-secured avionics data link defending against semantic spoofing attacks using a Physics-Informed Digital Twin (PIDT) validator.
Showcases the closed-loop PHM Digital Twin for the Boeing 787-8 ignition system, featuring deep feature spoofing and FastAPI microservice integration.
High-fidelity multibody simulation showcasing CS-23.473 hard-landing compliance validation and dual-input LSTM structural stress estimation.
Integrates a Grafana dashboard email alert system to notify MRO officials immediately when an abnormal landing occurs. Provides an early warning system derived from landing gear architecture telemetry.
Unified pipeline simulation tracking ArduPilot SITL flight physics, FlightGear visualization, and YOLOv8 aerial detection under a synchronized MAVLink log.
Visualizing the 4.3M-parameter generative Conditional Variational Autoencoder augmenting scarce RDE CFD data to generate physically consistent spatiotemporal fields.
Demonstrates the gas turbine low-pressure and high-pressure blade health monitoring system, tracking degradation and classifying micro-fractures.
Live inference of the neural network surrogate trained on airfoil sweep data to predict lift and drag coefficients in real-time.
Interactive TPE331‑25 digital twin demonstrating real‑time fault diagnosis, trend analysis, RUL estimation, and certified maintenance reports.