AiBased Deep Learning Technique for Electricity Demand and Peak Load Forecasting

Project Code :TCMAPY2330

Objective

The objective of this project is to create an AI-based deep learning system for accurately forecasting electricity demand and peak load using advanced models like Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). The project aims to address the growing need for efficient energy management by leveraging historical data that includes variables such as Hour, Day of the Week, Temperature, Humidity, Wind Speed, Industrial Load, and Residential Load. By capturing the temporal dependencies in this data, the system will provide precise predictions for future electricity demand, allowing for better energy optimization. The project also seeks to develop a user-friendly web application using Python and Flask, making the forecasting models accessible to stakeholders for real-time predictions. The objective is to create a scalable and efficient solution that empowers energy managers and decision-makers with actionable insights, ultimately enhancing energy efficiency and supporting smarter energy consumption practices.

Abstract

Power theft is a significant issue for electrical utilities worldwide, leading to substantial financial losses and operational inefficiencies. Traditional methods of detecting power theft, such as manual inspections and simple sensor-based systems, are slow, inaccurate, and require extensive human resources. This project aims to develop an AI-based power theft detection system using machine learning (ML) techniques to automate and enhance the process of identifying irregularities in electrical usage patterns. The system leverages three machine learning models—Support Vector Machine (SVM), Random Forest (RF), and XGBoost—trained on a dataset containing key electrical parameters such as voltage, current, power factor, energy consumption, and meter readings. By analyzing these parameters, the models are capable of distinguishing between normal and suspicious behavior indicative of power theft. The project achieves high accuracy with XGBoost showing 97.95% test accuracy, followed by RF with 98.10%, and SVM with 97.50%. These models are integrated into a web application developed using Flask, offering a user-friendly interface for utility providers to monitor real-time data and receive alerts for potential theft incidents. This AI-based solution enhances the speed, accuracy, and scalability of power theft detection, providing a cost-effective alternative to traditional manual inspection methods and reducing operational costs for utilities.

Keywords: Power theft detection, Machine learning, XGBoost, Random Forest, Support Vector Machine, Electrical parameters, Anomaly detection, AI, Flask, Real-time monitoring.

NOTE: Without the concern of our team, please don't submit to the college. This Abstract varies based on student requirements.