关键词: 深度学习, 工业加热炉, 温度场, 软测量, CFD仿真, 红外图像, 模型预测, 网络训练

Abstract:

For easy coking furnace, due to the unstable combustion process, it will cause local overheating of the furnace tube, resulting in loss and damage of the furnace, so it is necessary to measure the temperature at various locations of the heating furnace. In this paper, a soft measurement method of temperature field of furnace tube in heating furnace is proposed based on infrared images and computational fluid dynamics. By selecting data from different operating conditions to construct an experimental dataset, a real-time prediction model of temperature field is trained by deep learning to realize the purpose that the model can output the corresponding temperature field when input working conditions, and the soft measurement of temperature field of industrial heating furnace is realized. The temperature field calculated through CFD simulation in this study has an error of less than 5% compared to the temperature values on the thermocouple on the furnace wall. Therefore, it can be considered that the three-dimensional temperature field of the heating furnace obtained from CFD simulation can well represent the real temperature field inside the furnace.

Key words: deep learning, industrial heating furnace, temperature field, soft measurement, CFD simulation, infrared images, model prediction, network training,