IMPLEMENTATION OF STANDARD SCALER PREPROCESSING AND SUPPORT VECTOR MACHINE (SVM) ALGORITHM FOR DISAGREGATION BASED ON INTERNET OF THINGS (IOT)

• Main Author: Ermasari, Nofri
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55122
• Institution: Institut Teknologi Bandung

World energy consumption always increases every year, but world energy availability always decreases. World energy is like fossil energy which cannot be renewed and will decrease due to continuous exploration, which will cause an energy crisis. Therefore, it is necessary to make efforts to overcome the energy crisis, one of which is by saving energy, namely carrying out energy management by involving users. Users can take advantage of the power consumption data to make energy savings by using smart meters. In this research, a monitoring system for household electrical energy consumption with the energy disaggregation method based on the Internet of Things (IoT) has been built. This study aims to monitor loads and determine the electrical energy consumption of household appliances. The model validity test was carried out on five incandescent lamp devices connected to the kWh meter PZEM-004T and the Wifi module ESP8266. The kWh meter measures the power of five incandescent lamps at a time and will be read by the reader module. After that the power data will be sent to the server address with the MQTT protocol and sent to Node-Red to be stored in the database in realtime. Furthermore, the data is taken from the database to be processed using a disaggregation algorithm. The machine learning model that was built was the standard scaller preprocessing and the Support Vector Machine (SVM) algorithm with the input power and time parameters, the light state output parameters using a linear kernel. The model that was built succeeded in disaggregating energy with a simple load, namely predicting lamp activity with 100% accuracy. Then the model is applied to disaggregate household appliances contained in the AMPds dataset. In the AMPds dataset, data preprocessing is carried out to determine the state of each device that is the parameter of the model output using the K-Means algorithm. After that, the load disaggregation was carried out on household equipment, which varied in number, namely 5, 6 and 7 devices, obtaining an accuracy of 96.44%, 94.15%, and 91.82%. Furthermore, analysis of energy consumption for 30 days is carried out to determine the cost structure of household appliances used.