User-level Individual Residential Load Forecasting and Load Analysis for Residential Demand Response in India

Abstract

Peak demand in India is growing at such a rate that current capacity may not be sufficient to meet it in nearby future. A cost-effective solution for peak demand reduction in India is Residential Demand Response (RDR). Individual meter-level load forecasting is crucial for many smart grid applications like Demand Response management, by letting the customers to understand their own energy consumption ahead and change their consumption accordingly in case of time varying prices. This paper focuses on individual house electricity forecasting on half-hourly granularity for Indian residential customers and RDR. To achieve this objective, in this work, a popular machine learning model Support Vector Regression (SVR) is used. The half-hourly consumption data has been created for 9 houses for a year based on their consumption patterns. Load analysis for this forecasted data during on-peak hours at individual and aggregate level is also done in this paper. Our work demonstrates how individual meter-level load forecasting and load analysis of forecasted data can be used to help customer and utility to encourage price-based Demand Response programs at residential level and hence to achieve peak demand reduction and good savings through customer awareness. As Indian electric sector is working towards smart meter installations and time varying tariff in residential sector, this proposed work may help the residential customers and utility to reap benefits in future through programs like Demand Response.