Study on Quantitative Prediction Scheme of Aircraft Icing Based on Random Forest Algorithm

In this paper, a new aircraft icing prediction scheme is proposed to obtain the aircraft icing shape from common meteorological parameter. Machine learning modeling is used to establish the mapping between meteorological parameters and in-cloud microphysical parameters based on a random forest algorithm. The outputs of machine learning model, median volume diameter (MVD) and liquid water content (LWC), are utilized as input parameters to simulate ice accretion for a specific airfoil, and the final icing shape is determined. The present work shows that in-cloud microphysical parameters might have some relationship with common meteorological parameters, and random forest shows better performance in prediction of in-cloud microphysical parameters. The research work has brought about a quantitative prediction scheme of aircraft icing that shows high engineering practical value in route planning, aviation meteorological warning and airworthiness certification, etc.& COPY;2023 L & H Scientific Publishing, LLC. All rights reserved.