Virtual Sensor of Li-Ion Batteries in Electric Vehicles Using Data-Driven Analytic Thermal Solutions

Lithium-ion batteries, especially for electric vehicles (EVs), present safety risks, suffer poor performances, and undergo rapid degradation when operating under high temperatures. This, therefore, necessitates thermal monitoring for timely intervention. However, computations for this purpose can be very expensive and difficult to implement in real time. To overcome this problem, we establish a framework based on closed-form solutions to heat equations to estimate important parameters based on measurement data. They will be used for deducing heat generation rates for constructing forward-monitoring models for estimation. Our results show that the root-mean-square error between the estimated and actual temperature is at most 0.23 for sensor input interval between 50 and 60 s over the monitoring time of 1200 s, both with and without varying input currents. In addition, our proposed method achieves computations circa 350 times faster than that of finite element methods.