Generative adversarial network (GAN), as a powerful deep learning model, has demonstrated significant potential and application prospects in the field of infrared image simulation in recent years. This paper first summarizes the development of infrared simulation technology both domestically and internationally. It then details the latest progress and main methods of GAN-based infrared image simulation research, which are mainly divided into two methods: infrared simulation using random noise and infrared simulation using visible light images. It also describes the quality evaluation methods and evaluation indicators of generated infrared images (mainly subjective and objective evaluation methods). Finally, the future development trends of GAN in the field of infrared simulation are prospected.