Lithium polymer battery modelling and fault detection design

The accurate battery model and parameters identification are used to produce a reliable Battery Management System (BMS). In this research, the battery model using the equivalent circuit Thevenin model is proposed after considering its complexity, model accuracy, and robustness. Parameters identification is done by using pulse test data that contains current and Vd (the difference between Open Circuit Voltage (OCV) and terminal voltage) data that represent the battery characteristics. Recursive Least Square (RLS) algorithm is used to estimate the parameter recursively in order to lighten the computation process. The fault detection is also simulated using Matlab Simulink as a design of effective and efficient BMS to protect the battery from damage or failure. The results show that the battery modelling with the equivalent circuit Thevenin model can represent battery dynamic well. Parameters identification with the RLS algorithm shows accurate results with RMSE of 0,0021. The validation result also shows that the parameters obtained are accurate with the error of 0,0104%. The fault detection simulation also shows accurate detection toward any fault operation of the battery. It can detect faults in some parameters such as SOC fault, OCV fault, and overvoltage.