A STATE OF THE ART COMPOUND MODEL FOR CONSUMER TYPICAL LOAD PROFILE FORECASTING

Abstract

The rollout of advanced residential meters and sensors has been progressive over the years. This initiative is turning into better planning, forecasting, regulation and optimization of power systems. With the advent of Smart meters, consumer data analytics is becoming a prominent research area in which supervised clustering is one of the most common. Furthermore, predicting the typical load profile (TLP) of consumers is important for short, medium and long term load forecasting. With the increasing complexity, varying load and renewable energy interventions, forecasting demands for deep computational algorithms. This paper presents a State of the Art Compound model (SACM) to forecast TLP of electricity consumers. The model combines the fringe benefits of smart meters installed at the consumer’s premises and modern machine learning tools such as K-means Clustering, Principal Component Analysis (PCA) and Recurrent Neural Network (RNN). The effectiveness of the model is substantiated through data collected from smart meters in Mathura, Uttar Pradesh, India.