Implementation of artificial neural network for predicting water drag of the aircraft floater

Archipelago countries certainly need cheap and fast transportation facilities to run the economy of people who live separately from the sea. One solution is a modified small aircraft with a pair of floaters to float and glide in the waters. At the time of take-off, the aircraft encounters a water drag that will prevent it from accelerating and taking off. This research aims to predict the water drag of the floater using machine learning with an Artificial Neural Network algorithm (ANN). Prediction results are used for the analysis of the thrust required for the aircraft to take off. The ANN architecture consists of one hidden layer with seven neurons, one input layer of two source nodes, one output layer of seven neurons; the activation function used is sigmoid with Nadam's optimization method. ANN algorithm is run using python-libraries. Water drag calculation using the ANN model on three floaters with different dimensions shows that the predicted value is close to the target value obtained from the Savitsky method values. The regression coefficient is close to 1, with the error approaching 0. The use of machine learning ANN in calculating water drag has also proven to be 5-7 times faster than using numerical modeling via the Savitsky method, thus saving research time and costs.