Improving Temperature Prediction Precision for Aerial Cameras by Correcting Structure and Parameters of Thermal Network Models

Accurate temperature prediction is a prerequisite for precise thermal control of an aerial camera, which is crucial for capturing high-resolution images. This paper introduces a methodology for improving the precision of temperature predictions generated by thermal network models for aerial cameras, which involves correcting the structure and parameters of the thermal network. The passive thermal control optimization design of the frame insulation structure and convective heat transfer coefficients ensures that key thermal parameter values are maintained within optimum ranges. Following an analysis of the change trend consistency between the predicted temperatures and the transient thermal experimental data, the thermal network structure is modified to reduce the discrepancy between the thermal network and physical models. Using the Monte Carlo algorithm, parameter spaces for key thermal parameters are explored using different sampling numbers. The optimal parameter values are then derived by satisfying the tolerance limit condition and minimizing the objective function. As the sampling number increases, the RMSE between the predicted and experimental results decreases from 1.82 °C to 1.44 °C, and the R-squared value increases from 0.78 to 0.91. The proposed correction method demonstrates its efficiency and validity in refining the accuracy of thermal network models for aerial cameras.