A Prediction Method of Ship Motion Based on LSTM Neural Network with Variable Step-Variable Sampling Frequency Characteristics

In active heave compensation, in order to realize the smooth control of compensation platform, it is necessary use ship motion measurement system accurately obtain displacement signal, invert and then expansion contraction electric cylinder so that platform remains horizontal. The generally adopts second integral acceleration sensor signal. During acquisition process quadratic integration acceleration, communication output command, there a processing lag which makes error accumulate, resulting delay motion. collect more precisely, this paper proposes prediction method based on variable step-variable sampling frequency characteristic LSTM (Long Short-Term Memory) neural network. First, we autocorrelation function algorithm calculate inherent signal by system. Secondly, network used predict step lagging process, found difference will lead change delay—experiment laboratory verify. By controlling simulate using laser measure synchronously, verified does have an delay. Thirdly, sailing ship, signals are collected setting multiple sets systems. Finally, steps set, predicted It accuracy related collector network, improves model timeliness acquisition.