Research On Ship Track Forecast Based On AIS

Maritime transport plays an important role in the world.However,with the intensive traffic in many coastal and port waters,the situation is becoming more and more serious,and the frequent problems of maritime traffic accidents need to be solved urgently.Therefore,exploring ship navigation rules and track reliability prediction is of great significance to the safety of ship navigation,track planning and risk early warning.The automatic identification system information captured from the satellite can receive navigation information from the ship,such as position,heading,speed,destination,etc.,which provides an information basis for the ship’s tracking prediction and greatly enhances the maritime surveillance capability.This paper studies the ship track prediction based on the information of automatic identification system.The main contents include:(1)A ship track prediction algorithm based on optimized Back Propagation neural network is proposed.This algorithm addresses the randomness of the initial weights and thresholds of the Back Propagation neural network and is prone to fall into local optimization.It proposes to use genetic algorithm,particle swarm algorithm,ant colony algorithm,differential evolution algorithm and genetic particle swarm hybrid algorithm to optimize the Back Propagation neural network The initial weights and thresholds of the network make the training of the prediction model reach the state of global optimization.The simulation results show that the prediction accuracy of the optimized Back Propagation neural network algorithm has been significantly improved,and the optimized Back Propagation neural network prediction result of the hybrid genetic and particle swarm optimization algorithm is the best.(2)A ship track prediction algorithm based on an optimized adaptive network fuzzy inference system is proposed.This algorithm uses genetic algorithm and particle swarm algorithm to find the optimal Gaussian membership function parameters of the adaptive network fuzzy inference system model,and can solve the local optimization problem of the gradient adjustment part of the adaptive network fuzzy inference system.The simulation results show that the prediction accuracy of the optimized adaptive network fuzzy inference system has been greatly improved.The prediction accuracy of the adaptive network fuzzy inference system optimized by particle swarm optimization is higher,and the mean square error of its latitude and longitude prediction is 4.1109 e-6 and 3.858e-6.(3)A ship track prediction algorithm based on convolutional neural network and bidirectional long-term and short-term memory model is proposed.The algorithm uses the convolutional neural network to extract the potential characteristics of ship data,which improves the validity of the input data.The bidirectional long-term and short-term memory model is used to comprehensively consider the influence of past and future information,train the sample data,form the desired input-output mapping relationship,and then predict the future trajectory of the ship.Simulation experiments show that the prediction performance of the algorithm is better than the long-short-term memory model,and the delay estimates of the two are very similar.