Since the use of photovoltaics in the domestic and industrial sector is increasing, the reliability and safety of photovoltaic power generation systems is becoming more and more important. One of the crucial issues for safety is the formation of a hot spot, which can reach several hundred degrees Celsius in less than thirty minutes. For this reason, a fire may occur under prolonged exposure. However, a method for hot-spot detection during operation, based solely on current-voltage characteristics, has not yet been established because the detailed formation mechanism of these hot spots is different for every sample. We propose a new electro-thermal coupled model to predict the temperature distribution in a solar cell with a hot spot that is caused by a reverse bias voltage due to a local partial shade. We compare the results of the numerical simulations with experimental data and found that our coupled model can predict temperature distributions with an accuracy better than thirty degrees Celsius.