Development of an atmospheric boundary layer detection system based on a rotary-wing unmanned aerial vehicle.

To enhance meteorological detection methods, an atmospheric boundary layer detection system based on a rotary-wing unmanned aerial vehicle (UAV) was proposed. Computational fluid dynamics (CFD) was employed to model the surrounding airflow distribution during UAV hovering, thereby determining the optimal positions for sensor installation. A novel radiation shield was designed for the temperature sensor, offering both excellent radiation shielding and superior ventilation. To further improve temperature measurement accuracy, an error correction model based on CFD and neural network algorithms was designed. CFD was used to quantify the temperature measurement errors of the sensor under different environmental conditions. Subsequently, random forest and multilayer perceptron algorithms were employed to train and learn from the simulated temperature errors, resulting in the development of the error correction model. To validate the accuracy of the detection system, comparative experiments were conducted using the measurement values from the 076B temperature observation instrument as a reference. The experimental results indicate that the mean absolute error, root mean square error, and correlation coefficient between the experimental temperature errors and the algorithm-predicted errors are 0.055, 0.066, and 0.971 °C, respectively. The average error of the corrected temperature data is 0.05 °C, which shows substantial agreement with the reference temperature data. During UAV hovering, the average discrepancies between the temperature, humidity, and air pressure data of the detection system and the ground-based reference data are 0.6 °C, 1.6% RH, and 0.77 hPa, respectively.