Due to the solar radiation effect, current air temperature sensors may produce a radiation error that is on the order of 1 °C. In this paper, a temperature sensor with a new structure is proposed. Initially, a computational fluid dynamics (CFD) method is employed to quantify the radiation errors for the sensor under various environmental conditions. Then, a radiation error correction method is obtained by using a multi-layer perceptron (MLP) network. Finally, air temperature observation experiments are performed. The experimental results demonstrate that the mean radiation error is 0.07 °C. The mean absolute error (MAE), root mean square error (RMSE) and correlation coefficient (r) between the calculated radiation errors and the measured radiation errors are 0.027 °C, 0.034 °C, and 0.443, respectively.

由于太阳辐射效应，当前的空气温度传感器可能会产生大约 1°C 的辐射误差。在本文中，提出了一种具有新结构的温度传感器。最初，采用计算流体动力学 (CFD) 方法来量化传感器在各种环境条件下的辐射误差。然后，通过使用多层感知器（MLP）网络获得辐射误差校正方法。最后进行气温观测实验。实验结果表明，平均辐射误差为 0.07 °C。计算的辐射误差与测量的辐射误差之间的平均绝对误差 (MAE)、均方根误差 (RMSE) 和相关系数 (r) 分别为 0.027 °C、0.034 °C 和 0.443。