🌞 Solar PV Digital Twin — Advanced ML Dashboard & API

A production-grade machine learning system for Solar PV Power Output Prediction, complete with:

• 🤖 ML Training Pipeline (training_enhanced.py)
• 🌐 FastAPI Prediction Server (server.py)
• 🎨 Modern Streamlit Dashboard (app_enhanced.py)
• 📊 EDA + Model Visualizations
• 🧪 Batch CSV Prediction
• 🔴 Live Digital Twin Simulation
• 📁 Industry-level Project Structure

---

📸 Website UI — Screenshots

🟦 Dashboard Overview

🟩 Manual Predictor

🟧 Live Digital Twin Simulation

🟪 Batch CSV Predictions

---

📊 EDA & Model Visualizations

🔹 Feature Distributions

🔹 Correlation Heatmap

🔹 Pairplot (Top Features)

---

🧠 Project Summary

This Solar PV Digital Twin predicts instantaneous power output (W) using:

• Solar irradiance
• Temperature
• Wind speed
• Humidity
• Plane of array irradiance
• Cell temperature
• Panel tilt & azimuth

The system integrates ML + API + UI to simulate real-world solar plant monitoring, forecasting, and analysis.

---

🧮 Model Performance

Metric	Value
MAE	0.0 W
MAPE	4.93%
R² Score	0.9982
Max Error	0.05 W

Artifacts stored in:

models/pipeline_prod.joblib
models/metrics.json

---

🏗️ System Architecture

data/  → preprocessing → training_enhanced.py
          ↓
models/pipeline_prod.joblib
          ↓
FastAPI server (server.py)
          ↓
Streamlit Dashboard (app_enhanced.py)

---

🚀 Quick Start

1️⃣ Create Environment

python -m venv venv
.\venv\Scripts\Activate.ps1
pip install -r requirements.txt

2️⃣ Train Model

python src/training/training_enhanced.py

3️⃣ Start API Server

uvicorn src.server:app --reload

4️⃣ Launch Dashboard

streamlit run src/app_enhanced.py

---

🔌 API Endpoints

POST /predict

JSON → single power output prediction

POST /predict_batch

Upload CSV → returns full predictions CSV/JSON

GET /health

Health check

GET /model

Model metadata + metrics

---

📂 Project Structure

solar-pv-project/
│
├─ data/
├─ models/
│   ├─ pipeline_prod.joblib
│   ├─ metrics.json
│   └─ visualizations/
│
├─ src/
│   ├─ app_enhanced.py
│   ├─ server.py
│   ├─ training/
│   │   └─ training_enhanced.py
│   ├─ inference/
│   └─ utils/
│
├─ tests/
├─ requirements.txt
└─ README.md

---

🧩 Production Checklist

• ✔ Pre-commit formatting (Black + Flake8)
• ✔ Configurable training pipeline
• ✔ FastAPI endpoint validation
• ✔ Streamlit modular UI
• ✔ Ready for Docker deployment
• ✔ GitHub-friendly directory structure

---

👨‍💻 Author

Dhruv Bansal
ML Engineer • Dashboard Developer • Final Year EEE

---

📬 Contact

📧 Email: dhruvbansal23731@gmail.com 🌐 GitHub: https://github.com/dbansal0607

---