A Hybrid Intelligence Model Forecasts the Temperature of a Battery Used in Electric Vehicles

Battery technology is advancing due to the requirement to store reliable energy in electric transportation systems and portable electronics devices. Assessing battery conditions and increasing the autonomy of electric vehicles requires to improve the behavior models. Prognosticating battery performance, nevertheless, is complex due to multiple variables, such as temperature, wear, and usage guidelines. Understanding battery performance requires a depth knowledge of, for example, time series analysis, above all forecasting. This research uses Artificial Neural Networks (ANN) modeled taking into account system dynamics, to create a Lithium Ion Battery (LIB) prediction model. Testing was performed with a 59.2 VDC, 120 Ah LIB, monitoring voltage, current and temperature while simulating charges and discharges. The model used is a hybrid type, which uses clustering to generate internal local models. This model predict the temperature development with very good results in terms of Mean Squared Error. From this research it is deduced that hybrid models are appropriate tools for prognostic battery performance.