LSTM neural network-based ship position prediction method

The invention relates to a ship position prediction method based on an LSTM neural network. The method comprises the following steps: 1, constructing a ship position prediction model based on the LSTM neural network; 2, training and optimizing the constructed ship position prediction model; 3, collecting navigation information, control information and meteorological and hydrological information at the current moment; 4, judging whether the information is abnormal or not, if so, emptying the storage, and continuing data acquisition; 5, storing available data; 6, judging the available data volume, and if the available data volume is smaller than the set t group, continuing data acquisition; 7, t groups of continuous data including the current moment are reserved, and other data are emptied; 8, standardizing the reserved t groups of data; 9, inputting the standardized t groups of data into the ship position prediction model; and 10, outputting the predicted ship position at the target moment. According to the me