A Three-Layer Approach for Lithium-ion Battery State of Charge Estimation and Prediction using Deep Neural Networks: A Practical Implementation

Lithium-ion batteries play a pivotal role in stationary energy storage, serving as indispensable components for storing renewable energy and ensuring its availability when needed. Precisely estimating the state of charge (SOC) of these batteries is essential for efficient energy management. Addressing this challenge is critical to optimizing the performance of renewable energy systems. Neural networks present a solution for accurately estimating SOC without requiring an in-depth understanding of the battery's internal electrochemical state. This article presents a concept of a three-layer approach for lithium-ion battery modelling, SOC estimation and SOC prediction using a combination of deep neural networks (DNN) and recurrent neural network (RNN) with the gated recurrent unit (GRU) mechanism. The concept is demonstrated using battery discharge profiles from real electric vehicle driving cycles provided in IEEE Dataport database.