Short term load forecasting based on Internet of Things (IoT)

Abstract

In the era of internet every device is getting connected to the internet. In this paper, we have assumed that IoT devices can share their power consumption history. Based on data points collected from real world environment we have conducted experiments to show that IoT can be used as a reliable backbone of a short term load forecasting system. In the experiment four machine learning algorithms Long Short-term Memory (LSTM), Support Vector Machines Regression (SVR), Decision Forest Regression with AdaBoost and Nearest Neighbors Regression were used to analyze the performance of the load forecasting system. In the experiment Long Short Term Memory Network has given comparatively better result than other three machine learning algorithm with a root mean square error of 1.82.