Sulphur Hexafluoride (SF6) has been widely utilized in the Gas Insulated Switchgear (GIS) due to its great insulation ability. However, SF6 has great greenhouse effect. Liquid nitrogen (LN2) has been considered as a promising substitute for the SF6 gas because of its good insulation, arc quenching and cost effectiveness. To accurately simulate the LN2 switch, it is essential to establish a mathematical model for the LN2 arc. Black-box models have been commonly used in describing the dynamic characteristics of gas arc in circuit breaker simulations. There are numerous types of black-box models for gas arc, yet there is no literature available about the application of black-box model for the LN2 arc. This paper aims to establish the black-box model of LN2 arc. Based on the experimental data, several kinds of black-box models including Mayr, Cassie, Schwarz, Habedank and TP KEMA models were established, and their parameters were optimized by the Particle Swarm Optimization (PSO) algorithm. The performance of these black-box models was evaluated by the conductance error. The results indicated that black-box models can be employed for LN2 arc simulations, and the TP KEMA model exhibits the best performance with minimal conductance errors throughout the entire arcing process.