Abstract: The combustion process of industrial heating furnace is unstable and local over-temperature may occur at random locations, leading to furnace tube loss and damage. Since the heating furnace equipment is huge, it is difficult to measure the relevant physical quantity online. Therefore, this paper proposes to build a digital twin model based on the condition generation adversary-network on the standard industrial furnace heating model through which the temperature field of the heating furnace is predicted and the Unity 3D software is used for visualization to help the staff to optimize the combustion. The experimental results show that the cGAN network architecture can predict the temperature field, and the absolute error value of the temperature calculated at about 80% of the nodes within the test set is below 1% of the maximum temperature of the furnace, which is promising for the application.

Key words: online prediction, 3D temperature field, industrial heating furnace, cGAN, unity 3D visualization