Online battery temperature prediction is beneficial when it comes to taking effective heating and cooling actions in advance to realize better thermal management. First, this paper derives a simple thermal managing equation, i.e., a lightweight temperature prediction approach based on basic electrical and thermal laws from an equivalent circuit model. Then, to improve adaptability to complex conditions, some previously neglected thermal processes like heat conduction hysteresis and heat radiation are taken into consideration to achieve more realistic effects. In particular, the considerable reversible heat during alternating charging and discharging due to the electrochemical reactions of solid–liquid phase transitions is also modeled as a dependent variable in terms of real-time ohmic heat. A series of experiments is conducted under various loads and ambient conditions. The obtained results show that the proposed temperature prediction method can follow real temperature trajectories with an accumulated error of only 2.1 °C after an hour running.