As an essential equipment in ironmaking, formulating control strategies and ensuring the safety of the equipment on-site were facilitated by monitoring the temperature changes on the surface of the furnace shell for hot air furnace. In response to the limitations of thermal imaging systems, which have significant errors and are unable to monitor the entire area of the hot stove surface, contact temperature measurement was utilized for sensor arrangement optimization. The elbow rule and simulated annealing algorithm were employed to determine the optimal number and location of temperature sensors. By using the radial basis function method, the temperature field on the hot blast stove's surface was reconstructed before and after sensor optimization. The range of root-mean-square error values in different regions was then determined. The results indicate that compared to the temperature field with a uniform distribution of sensors, the range of the root mean square error value of the optimized temperature field changes from 0.02 to 0.165 to 0.02 to 0.145. This reduction not only narrows the range of the root mean square error value but also enhances the accuracy of the temperature field reconstruction, which can be used as a reference for the industrial temperature measurement of the hot air furnace.