Review and Analysis of Machine Learning Based Techniques for Load Forecasting in Smart Grid System

Abstract

Electrical load forecasting is an important process that can improve the efficiency and economy of the utility grid especially in the smart grid environment. Load forecasting plays a significant role in making decisions such as planning, generation scheduling, operation, pricing customer satisfaction, and system security. However, load forecasting is a tedious and difficult task due to the intermittent nature of Renewable Energy Systems (RES) that varies depending on the seasons and parameters such as change in temperature and humidity. Moreover, the connect loads are also complex in nature as they vary from season to season. Artificial Intelligent (AI) techniques are a promising approach for better load forecasting having chaotic and random variation of both load and generation. In this chapter, a load-forecasting algorithm for time series loads using AI techniques with supervised methods is presented and discussed. A comparative assessment of load forecasting based on supervised artificial intelligent algorithms, such as Artificial Neural Networks (ANN), Support Vector Machine (SVM), and Extreme Learning Machine (ELM), is performed on smart meter data. The results are presented and performance of the selected algorithms are analysed.