NILM- BASED SYSTEM FOR ENERGY MONITORING OF COMPOUND HOUSES IN GHANA

Abstract Compound housing is a housing arrangement that stemmed from cultural family living settings. Over the years, it has evolved to denote a group of different households that share a number of utilities including electricity. Electricity bills and payments are a concern in such living units due to the inability to split bills fairly. Previous work done to solve this are either expensive, or do not take into consideration the complexity of electricity connections in such houses. This thesis aims at verifying if the Feedforward Artificial Neural Network (ANN) can be used to split bills by implementing a Non-Intrusive Load Monitoring System for Compound Houses in Ghana. The proposed algorithm’s performance is compared to Hidden Markov Model (HMM) and its variants. It was discovered that the Feedforward ANN had higher overall performance than the HMM and its variants, and is efficient for splitting electricity bills in compound houses.