Comparison of the performances of different algorithms for estimating the degree of icing in aircraft

Aircraft icing is one of the biggest threats to aviation from past to present. Icing can cause accidents in aircraft, resulting in deaths and injuries, as well as have negative effects on the flight performance of aircraft. In order to eliminate or minimize these negative effects, anti-icing systems and de-icing systems have been proposed and these methods are being developed. These methods developed against icing are mostly operated and turned off by the pilot, depending on the degree of icing on the aircraft. This situation causes more aircraft accidents, including icing-related accidents and pilot-related accidents. In order to prevent pilot-induced errors, systems that turn on/off the de-icing/anti-icing systems according to the degree of icing risk have also been studied. In this study, seven different artificial intelligence algorithms that can directly operate the ice prevention/removal system as a result of estimating the degree of icing risk for a sample aircraft are tested. The boosting algorithm, which makes the estimation of the optimum icing risk degree with 88.59%, was first made in the simulink and called, and the icing degree was estimated at different input values. The scheme modeled in Simulink was embedded in the STM32F4 Discovery control card by adding waijung blocks, and the results of icing prediction were observed with the help of RGB LEDs. At the end of the study, a model that can be used in the prediction of icing in aircraft and that will enable automatic activation of anti-icing and anti-icing systems in aircraft has been proposed.