Ship Navigation Situation Estimation Based On Multi-source Data Fusion

When a ship is sailing at sea,its own situation is susceptible to the influence of complex Marine environment.In order to ensure the safety of the ship sailing at sea,how to estimate the situation of the ship sailing is very important.This paper integrates diversified ship information,realizes the safety assessment and situation prediction of ship navigation,provides auxiliary decision-making for navigators and provides decision-making support for the relevant regulatory departments,and can effectively guarantee the safety of ship navigation at sea and reduce the probability of accidents.The main contents of this paper are as follows:First,the present situation estimation model is studied,and it is determined that the main contents of ship navigation situation estimation include the analysis and acquisition of situation elements,the evaluation of ship navigation safety situation and the prediction of ship navigation situation.Several data fusion algorithms used to solve the situation estimation are studied and an algorithm based on neural network is determined to estimate the situation of ship navigation.This paper analyzes the factors affecting the navigation of ships from three aspects:human factors,ship factors and environment factors.Secondly,the safety situation assessment of ship navigation is studied.Compared with different safety assessment methods,Support Vector Machine(SVM)is selected as the ship navigation safety assessment model,the selection of SVM kernel function and the determination of important parameters will affect the evaluation accuracy of the model,an optimized SVM model is designed for safety evaluation,and the feasibility and superiority of the model are verified by simulation experiments.Thirdly,the situation prediction of ship navigation is studied.It is mainly to complete the prediction of the important features that represent the ship’s navigation situation,including position(longitude and latitude),sailing speed and heading.Obtain ship-related data to establish a data set for situation prediction.By studying three commonly used prediction models,Including BP(Back Propagation)neural network,Extreme Learning Machine(ELM)network model,Long Short-term Memory(LSTM)network model,by comparing the prediction effect of the three prediction models on the same data set,the navigation situation prediction model based on LSTM is determined.Finally,in order to improve the prediction effect of LSTM network model,sparrow search algorithm is introduced to optimize the hyperparameters of LSTM network model.The sparrow search algorithm is a kind of new intelligent optimization algorithm,the optimization precision and convergence speed is better than other intelligent algorithms,but also has the disadvantages of this kind of algorithm,including the uneven distribution of population,local development ability of the algorithm and global search ability not harmonious,for such problems,this paper puts forward the improved the sparrow search algorithm,By optimizing population initialization and sparrow position update,the algorithm performance is improved effectively.The simulation results show that the LSTM network model with sparrow search algorithm can significantly improve the prediction effect of ship navigation situation characteristics,which verifies the effectiveness of the algorithm.