Forecasting energy consumption in home energy management systems using machine learning method

Abstract

Reducing energy consumption by capital construction projects at all stages of their life cycle is an urgent task for the construction industry and the housing and communal complex. The article discusses ways to reduce energy costs in the operation of residential buildings. The aim of the study is to develop a methodology for predicting energy costs when using a home energy management system based on the machine learning method. All devices included in the "smart home" system are divided into three types, for each of them a method for calculating energy consumption is described. The algorithm of the home energy management system is to receive information from the energy supplier about their cost an hour in advance, calculate the energy consumption of all devices and predict energy consumption based on the reinforcement machine learning method. The effectiveness of the chosen method and the reliability of forecasting were evaluated by comparing the results with real costs for the selected time and calculating the average absolute error and the average absolute error in percent. The results of the study indicate the promise of using the method of machine learning with reinforcement to build a home energy management system based on forecasting energy consumption over time.