Arc fault is the most serious fault of oil-immersed transformer, seriously threatening the stable operation of the power grid and personnel safety. The gas generated by the arc cracking insulating oil is the main source of arcing fault pressure. Arc discharge affects the gas generation process by influencing the energy deposition, which in turn affects the calculation of fault pressure. The arc black box model is often used to describe the arc voltage in gas, while the arc in transformer oil is actually an arc in gas. Therefore, this paper establishes the black box model of arc in transformer oil and optimizes the parameters using the genetic algorithm. For the situation where the arc voltage varies greatly in different cycles, a black box model with segmented calculation is proposed, which realizes the accurate calculation of arc energy. The results of this paper are of great significance for the energy calculation in the case of large variations of arc voltage, and also play an important role in supporting the calculation and analysis of arcing fault pressure.