Prediction of electrical energy consumption for Internet of Things in disaggregated databases

Recently the increase of energy consumption has leveraged the development of solutions to save electricity. One of these solutions has been the creation of energy-saving policies based on energy forecasting in smart environments. The main idea behind this solution is that the residences are instrumented with sensor and actuator networks in order to monitor and manage energy consumption. This paper presents an analysis of global energy consumption forecasting in homes with aggregated and disaggregated databases and to achieve this goal some standard regression algorithms from the literature were used. It was observed that these type of algorithms are more effective when applied in disaggregated databases. The results with data collected from a real environment showed that the proposed approach is able to reduce the Root Mean Square Error (RMSE) up to 80%, compared to the global energy forecasting approach that uses aggregated databases.