Non-Intrusive Load Monitoring Based on Advanced-Boosting Algorithm

Energy efficiency has already become an integral part of achieving energy-saving goals, especially at home or commercial flat levels. An effective way to deal with it is through monitoring the electricity consumption of individual home appliances. As a result, non-intrusive load monitoring (NILM) study has attracted many researchers and engineers. However, many problems still hinder the NILM application, such as lack of accuracy and reliability. This study focuses on the NILM application via advanced boosting proposed method. Five home devices are used, including television, rice cooker, kettle, freezer, and wash dryer due to their high frequency of use in a daily basis. Stochastic gradient decent (SGD) and AdaBoost are combined where SGD algorithm is employed as a base classifier. Data imbalance handling technique, dimensionality reduction, and two different data scaling are first conducted for preprocessing data. The result shows that the proposed advanced boosting can perform excellently, outperforming other compared algorithms as benchmarks.