Model-based dynamic multi-parameter method for peak power estimation of lithium–ion batteries

Highlights: ► The Thevenin model is improved and a genetic algorithm is used to find the optimal model parameters. ► Three commonly used peak power estimation methods for lithium–ion batteries are compared. ► An online peak power estimation algorithm with multi-parameter constraints is proposed. ► The four peak power estimation methods are evaluated and compared based on a dynamic driving cycle. -- Abstract: A model-based dynamic multi-parameter method for peak power estimation is proposed for batteries and battery management systems (BMSs) used in hybrid electric vehicles (HEVs). The available power must be accurately calculated in order to not damage the battery by over charging or over discharging or by exceeding the designed current or power limit. A model-based dynamic multi-parameter method for peak power estimation of lithium–ion batteries is proposed to calculate the reliable available power in real time, and the design limits such as cell voltage, cell current, cell SoC, cell power are all used as its constraints; more importantly, the relaxation effect also is considered. Where, to improve the model's accuracy, the ohmic resistance of Thevenin model for the lithium–ion battery has been refined; in order to further improve the polarization parameters identification precision, a genetic algorithm has been used to gain the optimal time constant. Lastly, a test with several consecutive Federal Urban Driving Schedules (FUDSs) profiles is carried to evaluate the model-based dynamic multi-parameter method for peak power estimation. The experimental and simulation results indicate that the model-based dynamic multi-parameter method for peak power estimation can calculate the terminal voltage and the current available power much more reliably and accurately.