Designing a Neural Network Algorithm for Predicting Wave Height on the Surabaya - Banjarmasin Shipping Route

Abstract Indonesia is a maritime country, so maritime weather greatly affects marine activities, such as offshore and coastal building design, shipping, and fishing. In carrying out these activities, information about maritime weather is needed, especially wave height as a safety reference. Conventional wave height prediction is less able to describe sea conditions that have nonlinear properties, as a solution, the neural network algorithm can overcome the problem of nonlinear systems because of its ability to learn existing data. With a division of 80% training and 20% validation, this algorithm uses backpropagation as its learning method and multi-layer perceptron as a learning architecture. and produces an RMSE error value of 0.04. The smaller the RMSE, the more accurate the prediction. Thus, this algorithm can be used to predict wave heights on the Surabaya - Banjarmasin shipping route.