Improving Battery Model Accuracy Through Parameter Identification Using RLS and Pulse Test Analysis

Lithium batteries are an electrochemical-based electrical energy storage technology. The electricity generated comes from the chemical reaction of the positive and negative electrodes, which makes it complex and difficult to observe. To address this problem, models such as Electrical Circuit Models (ECM) are used, with the first-order Thevenin model being particularly accurate with key parameters Ri, RP, and Cp. The accuracy of this model depends on accurately determining these parameters. Traditional parameter identification methods generate only one set of values, despite the battery's parameter values changing with its state of charge (SoC). This research focuses on identifying battery parameters considering SoC changes, evaluated during a 5 SoC change from 100 to 10. To verify the impact on model accuracy, parameters were used to estimate open circuit voltage (OCV) under two test conditions: pulse test 1C and urban dynamometer driving schedule (UDDS). Parameter identification using the pulse test analysis (PTA) method with varying values resulted in more accurate OCV estimation, with mean absolute error (MAE) values of 4.072 × 10-3 at pulse test 1C and 6.395 × 10-3 at UDDS. © 2024 IEEE.