Robust xEV Battery State-of-Charge Estimator Design using Deep Neural Networks 2020-01-1181
The battery state-of-charge (SOC) is crucial information for the vehicle energy management system and must be accurately estimated to ensure more reliable and affordable electrified vehicles (xEV). However, due to the nonlinear, temperature, health, and SOC dependent behaviour of Li-ion batteries, SOC estimation is still a significant automotive engineering challenge. Traditional approaches to this problem, such as electrochemical models, usually require precise parameters and knowledge from the battery composition as well its physical response. In contrast, neural networks are a data-driven approach which requires no prior detailed knowledge of the battery or its nonlinear behaviour. The objective of this work is to walk through the design process to create a SOC estimator using deep feedforward neural networks (DNN). The process includes a description of how to obtain and prepare the data, and develop, tune and, validate a robust DNN capable of estimating Li-ion battery SOC despite sensors noise. To develop a robust estimator the DNN was exposed, during training, to datasets with errors intentionally added to the data, e.g. adding cell voltage variation of ±4mV, cell current variation of ±110mA, and temperature variation of ±5⁰C. The errors values were chosen to be similar to the noise and error obtained from real sensors used in commercially available xEVs. The robust DNN trained from two Li-ion cells datasets, NMC and NCA, is shown to overcome the added errors and obtain a SOC estimation accuracy of 1% root mean squared error (RMSE). The Matlab script and battery data will also be made available to download.
Carlos Vidal, Phillip Kollmeyer, Mina Naguib, Pawel Malysz, Oliver Gross, Ali Emadi
McMaster Automotive Res. Centre, FCA US LLC, McMaster University