Aircraft icing mechanism analysis methods

Abstract

To ensure flight safety, it is important to know how the icing processes of the aircraft aerodynamic surfaces occur. The article provides a review of works related to the analysis of the aircraft icing mechanism. According to publications, existing approaches to the analysis of the icing mechanism are divided into three groups: experimental research and testing, numerical modeling, and machine learning of neural networks. It is shown that experiments and tests give the most accurate results, since they are carried out in natural or close to natural flight conditions. Object-oriented results are obtained from numerical simulations when the input data set is tied to a specific aircraft. A disadvantage of numerical simulation is noted — a long calculation time. Attention is drawn to the fact that at present, machine learning methods for neural networks are being developed and are beginning to be implemented. These methods show a short computation time and predict not only the shape and size of ice, but also allow assessing the danger of icing and ranking the factors affecting icing, according to the degree of their importance. The article reveals the relationship of these three areas of analysis of the icing mechanism.