Independent energy supply in remote, rural areas with decentralized or localized energy systems (with solar photovoltaics/PV, for instance) saves costs of complex infrastructure such as for installation and maintenance of expensive power lines to remote villages, for instance in developing countries like Indonesia. Battery is the crucial component here without which, no localized energy systems would work by itself to provide stable, 24/7 power sources (Photovoltaics/PV only generates power in daytime, winds are very intermittent, etc.). Battery technology has advanced rapidly in the last decade or so, but these technologies will remain expensive for the application of localized, low-cost energy systems for remote and rural operations. The main objective is to design and develop a mini-scale smart, localized, independent but low-cost renewable energy systems as localized, reliable power producers for remote and marginalized operations (agriculture, maritime, military or disaster-response). We propose to use Machine Learning to first, predict the SOC (and/or the SOH) of legacy battery cells individually, and secondly, to optimize the charging of the individual cells accordingly at any given time, with varied temperature, charging/discharging cycles and other time-dependent parameters for a localized, self-sufficient energy system

Artificial intelligence, Battery, Smartdome, Machine learning, Optimization, Prediction.

• Christian Harito, S.T., Ph.D

• Dr. Muhammad Asrol, S.T.P., M.Si.

• Anak Agung Ngurah Perwira Redi, S.Kom., M.B.A, Ph.D.