Data mining model for evaluating and forecasting energy consumption by cloud computing

Abstract

According to high electrical energy consumption and rising energy costs, accurate model factory with a high performance is necessary to discover energy consumption patterns and forecast future demands. Factory sectors have a large share in global energy consumption; therefore, consuming energy in this section should be controlled and managed. In this study, a smart decision support system (SDSS) framework is applied in a cloud environment. It includes three main stages. The first stage collects data from a smart grid system and stores them in cloud databases. The second stage, which analyzes energy consumption data, is an analytic system including Autoregressive Integrated Moving Average (ARIMA) and Sensor Data Regularity-Tree (SDR-Tree) methods. The third stage is a web-based portal for user communication and displays the results on charts. Cloud computing technology presents services for a grid system infrastructure and software, which raises the speed and quality of processes and reduces the costs of storage devices. In the last stage, for speeding up the operations and reducing time response, a Load Balancing Decision Algorithm (LBDA) mechanism is applied in the cloud environment. The main aim of this study is to propose a model combined with two ARIMA and SDR-Tree methods in order to increase the accuracy of the results and solve the problems of both single models. Implementation of this hybrid model is suitable for the electrical energy efficiency improvement and smart factories development.