🚀 STLE Prediction in Hypersonic Vehicles

This project explores the use of supervised learning and deep learning models to predict Shock-Train Leading Edge (STLE) locations in hypersonic vehicles using sparse pressure sensor data.

📁 All code is implemented in Jupyter notebooks and based on experimental aerodynamic data.

---

🧠 Problem Statement

Given time-series pressure sensor readings from hypersonic wind tunnel experiments, the goal is to accurately predict the location of STLE – a key aerodynamic event that influences vehicle stability at high speeds.

---

🔍 Approach

• Built baseline and advanced supervised learning models:
  • Linear Regression
  • Random Forest
  • Support Vector Machines (SVM)
  • Feed-forward Neural Network
  • Recurrent Neural Network
• Engineered time-series features to denoise pressure signals and improve model accuracy
• Trained a deep neural network using only 7 of 37 total runs and evaluated on >900K test observations

---

Key Results

• 📈 Improved R² from 0.79 to 0.89
• 🧪 DNN generalized well to unseen test conditions, despite limited training set size