Machine learning approach for prediction of status of rechargeable batteries used in deep ocean moored buoy system using in-situ parameters

Battery packs are the only source of power storage of critical systems operated remotely at deep ocean location. The complete system operates only with the power available from the battery during the low sunlight condition. In this paper, the batteries used in oceanographic buoy systems operated in deep ocean regions far away from the mainland, types, backup mechanism along with the energy demand are briefed. Implementation of artificial neural network technique for understanding the status of battery systems are explained. An RMSE value of 0.29 is obtained between the predicted and actual value of battery voltage. This method will effectively indicate the status of the battery system of the buoy system and hence will provide decision-makers an idea on the operation and endurance of critical deep ocean buoy systems.