Data analytics for energy storage safety

Abstract

The lithium-ion battery has become the common type of rechargeable battery in consumer electronics. They are a well-known type of battery due to being one of the best energy-to-weight ratios, high open-circuit voltage, low self-discharge rate, no memory effect and a slow loss of charge when not in use [1]. Due to the high complexity in calculation cost of the lithium-ion battery being an electrochemical device, it creates a unique pattern caused by internal and external effects. Hence, this article proposes a method for accurately forecasting the remaining useful life (RUL) of lithium-ion batteries. This will prevent problems caused by continuous usage of the battery after reaching its life threshold. This paper presents the use of 2 algorithms to predict the RUL of the battery. The first proposed algorithm is using a Deep Neural Network (DNN) approach to train and predict the State of Health (SoH) of the battery. Next, the second proposed algorithm will be using a Long Short Term Memory type network (LSTM), to estimate the RUL of the battery. The proposed approach is implemented in a case study with a battery dataset obtained from the NASAAmes Prognostic Center of Excellence (PCoE) database. The results revealed that the performance accuracy is better with the second algorithms working together with the first algorithms. In addition, the result was capable of reaching the battery end of life cycle earlier than the actual battery.