Machine Learning-Based Aviation Meteorological Risk Prediction Model

Airlines prioritize passenger comfort and safety, and turbulence management (TM) is a vital component of aircraft operation. To promote secure and pleasant aero plane operations, this research presents an innovative machine learning-based technique for forecasting the presence of turbulence. We address the problem of anticipating uncommon turbulence events by constructing a specialized risk cluster based on past turbulence observations, utilizing combined meteorological data and turbulence observations. To produce a projection matrix for improved data representation, we used Kernel principal component analysis (KPCA) to minimize the dimension of the meteorological information. The risk cluster was then located by applying the fuzzy c-means (FCM) clustering technique. This was followed by the use of this novel fine-grained tree-integrated support vector machine (FGT-SVM) in conjunction with meteorological data from 2019 to anticipate turbulence-risk days. Our findings confirm the effectiveness of our machine learning model in forecasting turbulence occurrences based on intricate meteorological information by demonstrating a high accuracy in recognizing days with turbulence hazards. The experiment is run on a Python tool, and measures like, precision (89.45%), recall (85.75%), F1-score (90.76%) and G-mean (92.42%) are used to assess how well the study performed. Furthermore, by effectively employing sophisticated data analytics and machine learning techniques, this novel strategy not only promotes the safe operation of aircraft but also opens the door for improving aviation safety standards and human resource development.