☀️ Yulara Solar Digital Twin

A cloud-based digital twin platform for the Yulara Solar Farm (Northern Territory, Australia).

This system integrates:

• Machine learning forecasting
• Electricity price prediction
• Anomaly detection
• Interactive 3D visualization

into a scalable cloud architecture deployed on AWS and Netlify.

The platform demonstrates how AI + cloud infrastructure can optimize renewable energy operations through predictive analytics and real-time monitoring.

---

🚀 Live Deployment

Service	URL
Frontend (Netlify)	https://musical-tapioca-408587.netlify.app
Backend API (AWS Elastic Beanstalk)	http://yulara-backend-env.eba-mt7aim3j.us-east-1.elasticbeanstalk.com

---

⚙️ System Architecture

The platform follows a data ? ML ? API ? visualization pipeline. DATA INGESTION (AEMO NEM CSV + Solar Farm + Weather Data) ¦ DATA PIPELINE preprocess.py feature engineering yulara_master.csv ¦ ML TRAINING train_models.py

• Prophet ? Solar power forecasting • XGBoost ? Electricity price prediction • Isolation Forest ? Anomaly detection ¦ BACKEND — Flask REST API (AWS Elastic Beanstalk)

GET /api/stats GET /api/alerts POST /api/forecast/prophet POST /api/forecast/price POST /api/anomalies POST /api/historical POST /api/simulation/3d POST /api/predict/revenue ¦ FRONTEND — React Dashboard (Netlify)

• Power Forecast Dashboard • Price Forecast Dashboard • Anomaly Detection Panel • 3D Solar Farm Simulation • Revenue Prediction Tool

⚙️ Repository Structure

solar-digital-twin/

app.py Flask REST API serving all /api endpoints

preprocess.py Data preprocessing and feature engineering pipeline

train_models.py Trains ML models and stores artifacts

download_price_data.py Downloads electricity price data from AEMO

image_gen.py Generates model evaluation plots

requirements.txt Python dependencies

models/ best_power_model.pkl best_price_model.pkl isolation_forest_model.pkl anomaly_scaler.pkl

power_metrics.json price_metrics.json anomaly_metrics.json revenue_config.json training_summary.json

yulara-frontend/

src/ api/ index.js components/ App.jsx

public/ _redirects

package.json

---

📈 Machine Learning Model Performance

Model	Task	R² Score	Accuracy	MAE
Prophet	Solar Power Forecast	0.9805	98.05%	24.33 kW
XGBoost	Electricity Price Prediction	0.8986	89.86%	14.73 $/MWh
Isolation Forest	Anomaly Detection	0.97	97%	~3% anomaly rate

---

💻 Local Setup

Prerequisites

• Python 3.10+
• Node.js 18+
• pip

---

Clone Repository

git clone https://github.com/HUNT-001/solar-digital-twin.git

cd solar-digital-twin

---

Backend Setup

pip install -r requirements.txt

python preprocess.py python train_models.py

python app.py

Backend runs at:

http://localhost:5000

---

Frontend Setup

cd yulara-frontend npm install npm start

Frontend runs at:

http://localhost:3000

---

🌐 API Reference

Method	Endpoint	Description	Example Payload
GET	/api/stats	Solar farm statistics	—
GET	/api/alerts	Active anomaly alerts	—
POST	/api/forecast/prophet	Solar power forecast	{ "hours": 24 }
POST	/api/forecast/price	Electricity price forecast	{ "hours": 24 }
POST	/api/anomalies	Retrieve anomaly records	{ "n_records": 100 }
POST	/api/historical	Historical power data	{ "hours": 168 }
POST	/api/simulation/3d	Data for 3D solar simulation	{ "hours": 24 }
POST	/api/predict/revenue	Revenue prediction	{ "power_kw": 500 }

---

🚀 Cloud Deployment

Backend — AWS Elastic Beanstalk

zip -r yulara-deploy.zip app.py requirements.txt models/ eb deploy

Frontend — Netlify

cd yulara-frontend npm run build

Upload the build/ folder to Netlify.

---

📁 Dataset Sources

Dataset	Source	Description
Solar Generation	Desert Gardens Solar Farm	15-minute power generation data
Weather Data	Bureau of Meteorology	Temperature, irradiance, humidity
Electricity Price	AEMO NEM	5-minute settlement prices

Raw datasets are not included due to size (~400MB).

---

🛠️ Technology Stack

Layer	Technology
Frontend	React, Plotly.js, Three.js
Backend	Python, Flask
ML Forecasting	Prophet
ML Price Prediction	XGBoost
ML Anomaly Detection	Isolation Forest
Cloud	AWS Elastic Beanstalk
Hosting	Netlify
Data Processing	Pandas, NumPy

---

📖 Project Context

This project demonstrates a cloud-based digital twin architecture for renewable energy infrastructure, integrating:

• Machine learning forecasting
• anomaly detection
• real-time APIs
• interactive visualization
• scalable cloud deployment

The system showcases how AI-driven analytics can improve monitoring and optimization of solar energy systems.