Abstract
We present approaches to predict dynamic ditching loads on aircraft fuselagesusing machine learning. The employed learning procedure is structured into twoparts, the reconstruction of the spatial loads using a convolutionalautoencoder (CAE) and the transient evolution of these loads in a subsequentpart. Different CAE strategies are assessed and combined with either longshort-term memory (LSTM) networks or Koopman-operator based methods to predictthe transient behaviour. The training data is compiled by an extension of themomentum method of von-Karman and Wagner and the rationale of the trainingapproach is briefly summarised. The application included refers to a full-scalefuselage of a DLR-D150 aircraft for a range of horizontal and vertical approachvelocities at 6{\deg} incidence. Results indicate a satisfactory level ofpredictive agreement for all four investigated surrogate models examined, withthe combination of an LSTM and a deep decoder CAE showing the best performance.