Net Power Forecasting for PV-Integrated Residential Buildings in Delhi

Project Overview

This project forecasts net power (PV generation minus residential load) for Delhi's residential sector using:

• Photovoltaic (PV) power generation simulation
• Load forecasting with LSTM and XGBoost models
• Time-series analysis of net power balance

Key Features

• Accurate, Delhi-specific PV simulation using pvlib
• Robust forecasting via LSTM and XGBOOST
• Clear visual comparisons of actual vs predicted load
• CSV output for further analysis
• Highly accurate

Research Paper

This project is based on the research paper:

"Net Power Forecasting for PV-Integrated Residential Buildings in Delhi"

📄 Full Paper: Research_Paper_ML_PV net power.pdf

Abstract

The paper investigates the forecasting of net power demand in residential buildings integrated with photovoltaic (PV) systems in Delhi, India. By combining solar power generation modeling with machine learning-based load forecasting, this research provides accurate predictions of the net power balance (PV generation minus consumption) to support grid planning and energy management.

Key Contributions

• Delhi-specific PV modeling using meteorological data and pvlib simulation
• Machine learning forecasting with LSTM and XGBoost algorithms
• Time-series analysis of residential energy consumption patterns
• Net power prediction for optimal grid integration and energy storage planning