Ship Trajectory Research Articles

Application of an Encoder–Decoder Model with Attention Mechanism for Trajectory Prediction Based on AIS Data: Case Studies from the Yangtze River of China and the Eastern Coast of the U.S

With the rapid growth of shipping volumes, ship navigation and path planning have attracted increased attention. To design navigation routes and avoid ship collisions, accurate ship trajectory prediction based on automatic identification system data is required. Therefore, this study developed an encoder–decoder learning model for ship trajectory prediction, to avoid ship collisions. The proposed model includes long short-term memory units and an attention mechanism. Long short-term memory can extract relationships between the historical trajectory of a ship and the current state of encountered ships. Simultaneously, the global attention mechanism in the proposed model can identify interactions between the output and input trajectory sequences, and a multi-head self-attention mechanism in the proposed model is used to learn the feature fusion representation between the input trajectory sequences. Six case studies of trajectory prediction for ship collision avoidance from the Yangtze River of China and the eastern coast of the U.S. were investigated and compared. The results showed that the average mean absolute errors of our model were much lower than those of the classical neural networks and other state-of-the-art models that included attention mechanisms.

Identification of Spoofing Ships from Automatic Identification System Data via Trajectory Segmentation and Isolation Forest

Outliers of ship trajectory from the Automatic Identification System (AIS) onboard a ship will affect the accuracy of maritime situation awareness, especially for a regular ship trajectory mixed with a spoofing ship, which has an unauthorized Maritime Mobile Service Identification code (MMSI) owned by a regular ship. As has been referred to in the literature, the trajectory of these spoofing ships would simply be removed, and more AIS data would be lost. The pre-processing of AIS data should aim to retain more information, which is more helpful in maritime situation awareness for the Maritime Safety Administration (MSA). Through trajectory feature mining, it has been found that there are obvious differences between the trajectory of a regular ship and that of a regular ship mixed with a spoofing ship, such as in terms of speed and distance between adjacent trajectory points. However, there can be a long update time interval in the results of severe missing trajectories of a ship, bringing challenges in terms of the identification of spoofing ships. In order to accurately divide the regular ship trajectory and spoofing ship trajectory, combined with trajectory segmentation by the update time interval threshold, the isolation forest was adopted in this work to train the labeled trajectory point of a regular ship mixed with a spoofing ship. The experimental results show that the average accuracy of the identification of spoofing ships using isolation forest is 88.4%, 91%, 93.1%, and 93.3%, corresponding to different trajectory segmentation by update time intervals (5 h, 10 h, 15 h, and 20 h). The research conducted in this study can almost eliminate the outliers of ship trajectory, and it also provides help for maritime situation awareness for the MSA.

Robust Learning with Noisy Ship Trajectories by Adaptive Noise Rate Estimation.

Ship trajectory classification is of great significance for shipping analysis and marine security governance. However, in order to cover up their illegal fishing or espionage activities, some illicit ships will forge the ship type information in the Automatic Identification System (AIS), and this label noise will significantly impact the algorithm's classification accuracy. Sample selection is a common and effective approach in the field of learning from noisy labels. However, most of the existing methods based on sample selection need to determine the noise rate of the data through prior means. To address these issues, we propose a noise rate adaptive learning mechanism that operates without prior conditions. This mechanism is integrated with the robust training paradigm JoCoR (joint training with co-regularization), giving rise to a noise rate adaptive learning robust training paradigm called A-JoCoR. Experimental results on real-world trajectories provided by the Danish Maritime Authority verified the effectiveness of A-JoCoR. It not only realizes the adaptive learning of the data noise rate during the training process, but also significantly improves the classification performance compared with the original method.

Ship Trajectory Prediction: An Integrated Approach Using ConvLSTM-Based Sequence-to-Sequence Model

Maritime transportation is one of the major contributors to the development of the global economy. To ensure its safety and reduce the occurrence of a maritime accident, intelligent maritime monitoring and ship behavior identification have been drawing much attention from industry and academia, among which, the accurate prediction of ship trajectory is one of the key questions. This paper proposed a trajectory prediction model integrating the Convolutional LSTM (ConvLSTM) and Sequence to Sequence (Seq2Seq) models to facilitate simultaneous extraction of temporal and spatial features of ship trajectories, thereby enhancing the accuracy of prediction. Firstly, the trajectories are preprocessed using kinematic-based anomaly removal and Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) to improve the data quality for the training process of trajectory prediction. Secondly, the ConvLSTM-based Seq2seq model is designed to extract temporal and spatial features of the ship trajectory and improve the performance of long-time prediction. Finally, by using real AIS data, the proposed model is compared with the Seq2Seq and Bidirectional LSTM based on attention mechanism (Bi-Attention-LSTM) models to verify its effectiveness. The experimental results demonstrate that the proposed model achieves excellent performance in predicting turning trajectories, good predictive accuracy on straight line motions, and greater improvement in prediction accuracy compared to the other two benchmark models. Overall, the proposed model represents a promising contribution to improving ship trajectory prediction accuracy and may enhance the safety and quality of ship navigation in complex and volatile marine environments.

Nonparametric modeling of ship maneuvering motions in calm water and regular waves based on R-LSTM hybrid method

It is essential to establish reliable method to predict the ship maneuvering motions in complex navigation environments for navigation efficiency and safety. In this paper, a novel hybrid nonparametric modeling method (R-LSTM) is developed by coupling Bidirectional Long Short-Term Memory (Bi-LSTM) with residual network and attention mechanism. Firstly, the prediction capability of the proposed hybrid method is evaluated in the predictions of zigzag maneuvers in calm water. The model test data of a KVLCC2 tanker model are utilized for modeling and validation, and the optimal parameters are determined by Tree-structured Parzen Estimator algorithm. The results demonstrate that the proposed method has higher prediction accuracy and better generalization ability than the classical LSTM method. On this basis, the online modeling method for predicting the turning motion in regular waves is developed along with sliding time window and time series split cross-validation. The model test data of a KCS container ship and a ONRT tumblehome ship under different wave conditions are used for online modeling and real-time prediction. The results indicate that the ship trajectories with trajectory drift and fluctuation features during the turning motion under the wave impacts are captured well by the developed method with satisfactory confidence intervals.

Research into Ship Trajectory Prediction Based on An Improved LSTM Network

The establishment of ship trajectory prediction is critical in analyzing trajectory data. It serves as a critical reference point for identifying abnormal behavior and potential collision risks for ships. Accurate and real-time ship trajectory prediction is essential during navigation. Since the timing of automatic identification system (AIS) data is irregular, traditional methods usually use time calibration to simulate the data of uniform sequencing before analysis. Inevitably, this increases the chances of error and time delays. To address this issue, we propose a time-aware LSTM (T-LSTM) single-ship trajectory model combined with the generative adversarial network (GAN) to predict multiple ship trajectories. These analysis methods are capable of directly analyzing AIS data and have demonstrated better performance in both single-ship and multi-ship trajectories. Our experimental results show that the proposed method achieves high accuracy and can meet the practical navigation requirements of ships.

Research on the carbon emissions traceability inventory and multi-horizon prediction of ship carbon emissions: a case study of Tianjin Port

IntroductionIn recent years, the adverse effects of escalating maritime trade and international shipping– particularly in regard to increased greenhouse gas emissions and their impact on human health– have come to the fore. These issues have thus instigated a surge in pressure to enhance the regulation of shipborne carbon emissions.MethodsThe study utilized the automatic identification system (AIS) data, Lloyd’s register data, and pollutant emission parameters to calculate the carbon emissions from the main engine, auxiliary engine, and boiler of vessels under varying sailing conditions, utilizing the dynamic method of ships. In relation to geographic information and ship trajectory, a comprehensive inventory of ship carbon emissions was developed, revealing pronounced spatiotemporal characteristics. To assure the accuracy of the substantial AIS dataset, procedures including data cleaning, trajectory integration, data fusion, and completion were executed. Such processes are indispensable, given the potential for transmission and storage errors associated with AIS data. To forecast CO2 emissions over diverse time intervals, a temporal fusion transformer model equipped with attention mechanisms was employed.ResultThe paper furnishes a case study on Tianjin Port, wherein a high-resolution carbon emissions inventory was devised based on AIS data acquired from vessels. This inventory was subsequently employed to generate multi-feature predictions of future carbon emissions. Given the optimal parameter configuration, the proposed method attained P50 and P90 values of 0.244 and 0.118 respectively, thereby demonstrating its efficacy.DiscussionRecognizing the sources of ship carbon emissions in this region and forecasting such emissions in the future substantiates that this method accurately portrays the laws of ship carbon emissions. Our study provides a scientific basis for decision-making in port and pollution management, enabling the creation of targeted emission reduction policies for ships.

A full-time deep learning-based alert approach for bridge–ship collision using visible spectrum and thermal infrared cameras

With the rapid development of transportation infrastructure and the growing water transportation industry, many bridges have been constructed across navigable rivers. However, collisions between ships and bridges remain a frequent occurrence, particularly under low illumination conditions. To address this issue, this paper proposes a full-time bridge–ship collision alert approach based on the combination of visible spectrum and infrared cameras. This approach includes the following steps: (1) capturing and identifying ships in the channel using a single visible spectrum camera is challenging in low illumination environments such as nighttime. To overcome this limitation, we propose the use of infrared thermal imaging technology, which can monitor ships at night due to their inherent emissivity of ship thermal radiation. By comprehensively analyzing factors such as the real-time lighting environment and comparing the advantages of visible spectrum cameras and thermal infrared cameras in object recognition under various illuminance levels, we determine the effectiveness of the decision-making method of the dual-channel camera for ships under different illumination conditions. (2) To detect and track ships in the channel, a deep learning-based automatic detection and tracking method is designed using visible spectrum and infrared thermal imaging cameras, providing early warning of dangerous trajectories of ships that threaten the piers. To address the real-time identification and positioning of ships by dual-channel cameras in different light environments, we adopt a target detection network and a displacement calculation algorithm for multi-target tracking. Analysis results demonstrate that this method can automatically assign visible spectrum and thermal imaging cameras as effective input sources in different illumination environments to detect and warn ships with dangerous trajectories, contributing to the safety of navigable ships and bridges in low illumination conditions.

Ship Trajectory Prediction Based on the TTCN-Attention-GRU Model

As shipping continues to play an increasingly important role in world trade, there are consequently a large number of ships at sea at any given time, posing a risk to maritime traffic safety. Therefore, the tracking and monitoring of ships at sea has gradually attracted the attention of scholars. Ship trajectory prediction comprises an important aspect of ship tracking and monitoring. Trajectory prediction describes the forecasting of a ship’s future trajectory over a period of time through use of historical trajectory information of the ship, so as to predict the sailing dynamics of the ship in advance. Accurate trajectory prediction can help maritime regulatory authorities improve supervision efficiency and reduce collisions between ships. Temporal Convolutional Network (TCN) offers good time memory ability and has shown better performance in time series prediction in recent years. Ship trajectory sequence belongs to the category of time series. Thus, in this paper, we introduce TCN into the field of ship trajectory prediction and improve on it, and propose Tiered-TCN (TTCN). The attention mechanism is a way to help neural networks learn data features by highlighting features that have a greater impact on predicted values. Gate Recurrent Unit (GRU) is an important variant of Recurrent Neural Networks (RNN), which bears a strong nonlinear fitting ability. In this paper, TTCN, attention mechanism and GRU network are integrated to construct a hybrid model for trajectory prediction, which is referred to as TTCN-Attention-GRU (TTAG). By optimizing the advantages of each module, the prediction effect is achieved with high precision. The experimental results show that the TTAG model is superior to all the baseline models presented in this paper.

A novel method for ship trajectory prediction in complex scenarios based on spatio-temporal features extraction of AIS data

Ship trajectory prediction plays a key role in the early warning and safety of maritime navigation. Ship pilots must have a complete grasp of the future trajectories of ships within a certain period of time when moving the ship to effectively avoid collisions. However, the accuracy of ship trajectory prediction is a significant issue that needs to be resolved at present. In this paper, we propose a ship trajectory prediction model covering spatio-temporal awareness graph attention network (GAT) based on long short-term memory network (LSTM) to predict the future trajectories of ships in complex scenarios, named STPGL model, which adopts the design of encoder–decoder structure. The historical encoder uses an LSTM to extract the kinematic sequence features of each ship from historical trajectories in the temporal dimension. The interactive encoder summarizes the interaction features between different ships through GAT in the spatial dimension. Then, the two features are fused and fed into the decoder module to infer the future trajectories of ships. The experimental results demonstrate that STPGL model can effectively improve the prediction accuracy of short-term, medium-term and long-term ship trajectory. It has excellent performance and has a certain reference value for the advancement of unmanned ship collision avoidance.

“Ship-port-country” multi-dimensional research on the fine analysis of China's LNG trade

Liquefied Natural Gas (LNG) is a clean and low-carbon energy source that plays an important role in combating global warming. As the world's largest greenhouse gas (GHG) emitter, China has taken coal-to-gas as an important measure to reduce GHG emissions, and has become the world's largest LNG importer. Studying the patterns and characteristics of China's LNG trade is of great significance for ensuring energy supply security, formulating and adjusting energy strategies. However, the refined analysis of China's LNG trade is still inadequate, mainly due to the lack of research on the details of ship trajectories and port attributes. Therefore, a “ship-port-country” multidimensional LNG trade analysis method is proposed to improve the timeliness, multi-dimensionality and fineness of China's LNG trade analysis. The multidimensional characteristics of China's LNG trade are analyzed by studying the big data of nearly 3.5 billion automatic identification system (AIS) records in 2017. The study has found that (1) The arrival frequency of China's LNG trading ships varies greatly with time. For example, the most frequent daily arrivals are between 04:00 and 05:00 in the morning, while the least frequent arrivals are between 15:00 and 17:00 in the afternoon. (2) The Chinese LNG trade network consisting of 79 ports has been formed, of which NANTONG Port is the most important port. (3) The LNG trade volume calculated in this study has shown a strong correlation with the national LNG trade statistics and can be used to analyze LNG trade with higher spatial and temporal resolution. According to the characteristics of LNG trade at the three levels of ship, port and country, China can optimize its LNG trade policy from the following aspects: (1) Dynamically allocate port resources according to the arrival frequency of LNG tankers to reduce port congestion and waste of resources. (2) Expand core LNG trading ports such as NANTONG Port, ZHOUSHAN Port, TANGSHAN (JINGTANG) Port, XIUYU Port, DALIAN Port and LUSHUN Port. (3) Actively realize the diversification of LNG trading partners to reduce the dependence of LNG import sources on one country or region, and ensure energy security. The “ship-port-country” multi-level LNG trade analysis method proposed in this study provides a new research perspective for the analysis of the LNG trade pattern. This study reveals the multidimensional characteristics of China's LNG trade, which can provide decision-making support for China to formulate reasonable LNG trade policies and energy security policies, and is conducive to the stable and sustainable development of global LNG trade.

AIS data-driven ship trajectory prediction modelling and analysis based on machine learning and deep learning methods

Maritime transport faces new safety challenges in an increasingly complex traffic environment caused by large-scale and high-speed ships, particularly with the introduction of intelligent and autonomous ships. It is evident that Automatic Identification System (AIS) data-driven ship trajectory prediction can effectively aid in identifying abnormal ship behaviours and reducing maritime risks such as collision, stranding, and contact. Furthermore, trajectory prediction is widely recognised as one of the critical technologies for realising safe autonomous navigation. The prediction methods and their performance are the key factors for future safe and automatic shipping. Currently, ship trajectory prediction lacks the real performance measurement and analysis of different algorithms, including classical machine learning and emerging deep learning methods. This paper aims to systematically analyse the performance of ship trajectory prediction methods and pioneer experimental tests to reveal their advantages and disadvantages as well as fitness in different scenarios involving complicated systems. To do so, five machine learning methods (i.e., Kalman Filter (KF), Support Vector Progression (SVR), Back Propagation network (BP), Gaussian Process Regression (GPR), and Random Forest (RF)) and seven deep learning methods (i.e., Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gate Recurrent Unit (GRU), Bi-directional Long Short-Term Memory (Bi-LSTM), Sequence to Sequence (Seq2seq), Bi-directional Gate Recurrent Unit (Bi-GRU), and Transformer) are first extracted from the state-of-the-art literature review and then employed to implement the trajectory prediction and compare their prediction performance in the real world. Three AIS datasets are collected from the waters of representative traffic features, including a normal channel (i.e., the Chengshan Jiao Promontory), complex traffic (i.e., the Zhoushan Archipelago), and a port area (i.e., Caofeidian port). They are selected to test and analyse the performance of all twelve methods based on six evaluation indexes and explore the characteristics and effectiveness of the twelve trajectory prediction methods in detail. The experimental results provide a novel perspective, comparison, and benchmark for ship trajectory prediction research, which not only demonstrates the fitness of each method in different maritime traffic scenarios, but also makes significant contributions to maritime safety and autonomous shipping development.

The Study of Fishing Vessel Behavior Identification Based on AIS Data: A Case Study of the East China Sea

The goal of this paper is to strengthen the supervision of fishing behavior in the East China Sea and effectively ensure the sustainable development of fishery resources. Based on AIS data, this paper analyzes three types of fishing boats (purse seine operation, gill net operation and trawl operation) and uses the cubic spline interpolation algorithm to optimize the ship trajectory and construct high-dimensional features. It proposes a new coding method for fishing boat trajectory sequences. This method uses the Geohash algorithm to divide the East China Sea into grids and generate corresponding numbers. Then, the ship trajectory is mapped to the grid, the fishing boat trajectory points are associated with the divided grid, and the ship trajectory ID is extracted from the corresponding grid. The extracted complete trajectory sequence passes through the CBOW (continuous bag of words) model, and the correlation of trajectory points is fully learned. Finally, the fishing boat trajectory is converted from coordinate sequence to trajectory vector, and the processed trajectory sequence is trained by the LightGBM algorithm. In order to obtain the optimal classification effect, the optimal superparameter combination is selected. We put forward a LightGBM algorithm based on the Bayesian optimization algorithm, and obtained the classification results of three kinds of fishing boats. The final result was evaluated using the F1_score. Experimental results show that the F1_score trained with the proposed trajectory vectorization method is the highest, with a training accuracy of 0.925. Compared to XgBoost and CatBoost, the F1_score increased by 1.8% and 1.2%, respectively. The results show that this algorithm demonstrates strong applicability and effectiveness in fishery area evaluations and is significant for strengthening fishery resource management.

A novel asymmetric barrier Lyapunov function-based fixed-time ship berthing control under multiple state constraints

The construction of an asymmetric barrier function with internal security set, while ensuring its invariance in the presence of interference, is a challenging task to achieve stability and security. In this paper, a translation transformation is used to construct a novel asymmetric Barrier Lyapunov Function. To the best of our knowledge, this is the first designed continuously differentiable barrier Lyapunov function (BLF) of this type. A new adaptive sliding mode control (SMC) method based on asymmetrical BLF is proposed for uncertain external disturbances. The robust control method can ensure the safety performance in the presence of model uncertainty with bounded tracking error of the ship trajectory and velocity. Its fixed-time convergence is then proved by the Lyapunov theory. The results show that it has stronger anti-interference ability in bad sea conditions, better robustness, and concise form. In addition, the system can deal with unknown disturbances, ensure the safe navigation, and keep the good tracking of the route under wind and current disturbance due to random sea conditions.

МЕТОД ФОРМАЛІЗАЦІЇ ПРОЦЕСУ ПРИЙНЯТТЯ РІШЕННЯ ЩОДО ЗАПОБІГАННЯ НЕБЕЗПЕЧНИХ СИТУАЦІЙ В СИСТЕМІ E-NAVIGATION

The study aims to improve the safety of ship traffic by improving the methods of formalizing the decision-making process for preventing dangerous situations in a single e-navigation concept. The goal is achieved by a clear statement of the problem and the choice of appropriate approaches to its solution. The main stages of the decision-making cycle performed by the navigator in collision prevention are analysed. Particular attention is paid to the operations performed at the decisionmaking stage on the further movement of the vessel. The ship's movement plan is described by models of elementary behaviour, each of which consists of elementary actions represented by a set of fuzzy logic rules for performing a certain subtask. The structure of a system that implements the behavioural approach to decision-making by a navigator is proposed. Emphasis is placed on the need to coordinate vessels when deciding to plan a route based on models of elementary behaviours, and a method for coordinating the planning of the vessel's trajectory is developed. The sequence of stages of the method for coordinating the planning of the ship's trajectory is performed cyclically until the target point of the route is reached. At each stage of the vessel's movement, the navigation obstacles and other vessels entering the area of interest of the navigator are monitored and potential hazards from them are identified. A method for tracing the vessel's movement and existing vessels in the navigator’s interest area has been developed, which allows for predicting the location at subsequent moments and identifying the coordinates of the conflict site where a situation of dangerous proximity occurs. To formalize the areas dangerous for vessel traffic, a method for generating a map of navigational hazards and conflict situations in the navigator’s interest area has been developed, based on a modified method of potential fields. To formally describe the space, only the concept of a potential function is used to represent the space as a matrix of values. The most significant scientific result is the proposed approach to solving the problem of supporting the decision-making of a navigator based on a combination of methods using fuzzy logic and allowed domains, which allows obtaining options for actions that ensure the safety of ship movement in a dangerous situation. Keywords: ship, navigator, collision, navigation situation, navigation systems, e-navigation, tracing method, navigation hazard, navigation hazard map

A data mining method to extract traffic network for maritime transport management

Maritime traffic network is essential for navigation efficiency and safety of the maritime transport system. This study proposes a framework for extracting maritime traffic network based on Automatic Identification System (AIS) data. The framework consists of maritime traffic pattern recognition, semantic routes extraction, route decomposition, and network generation. Firstly, a data-driven method is introduced to recognize ship behavior patterns and extends the single ship behaviors to regional characteristics to determine the departure-arrival areas. Then, based on the different combination of departure-arrival areas, the ship trajectories are classified to traffic groups. Subsequently, the grid-system is used to rasterize each traffic group, which realizes the fusion of trajectory data and geographic location information. Finally, to obtain the main routes and navigation channels, the extraction method is introduced by establishing the cumulative grid importance function. The main routes, together with the navigation channels, compose the maritime traffic network. The method is applied to AIS data in the Beibu Gulf, and the results show that the traffic network contains 12 stop areas, 4 entry/exit locations, 13 main routes as well as their corresponding navigation channels. It is therefore concluded that the proposed method helps (1) provide a theoretical framework to obtain and analyze the maritime traffic network and (2) enrich navigation channel identification methods for maritime transport management.

Prediction of Ship Trajectory in Nearby Port Waters Based on Attention Mechanism Model

In recent years, the prediction of ship trajectory based on automatic identification system (AIS) data has become an important area of research. Among the existing studies, most focus on a single ship to extract features and train models for trajectory prediction. However, in a real situation, AIS contains a variety of ships and trajectories that need a general model to serve various cases. Therefore, in this paper, we include an attentional mechanism to train a multi-trajectory prediction model. There are three major processes in our model. Firstly, we improve the traditional density-based spatial clustering of applications with noise (DBSCAN) algorithm and apply it to trajectory clustering. According to the clustering process, ship trajectories can be automatically separated by groups. Secondly, we propose a feature extraction method based on a hierarchical clustering method for a trajectory group. According to the extraction process, typical trajectories can be obtained for individual groups. Thirdly, we propose a multi-trajectory prediction model based on an attentional mechanism. The proposed model was trained using typical trajectories and tested using original trajectories. In the experiments, we chose nearby port waters as the target, which contain various ships and trajectories, to validate our model. The experimental results show that the mean absolute errors (MAEs) of the model in longitude (°) and latitude (°) compared with the baseline methods were reduced by 8.69% and 6.12%.

Model predictive ship trajectory tracking system based on line of sight method

Model predictive ship trajectory tracking system based on line of sight method

A real-time multi-ship collision avoidance decision-making system for autonomous ships considering ship motion uncertainty

Ship autonomous collision avoidance has attracted increasing attention in recent years. However, more attention is paid to the scenario in which the target ship keeps its course and speed. Less attention is paid to the development of a collision avoidance decision-making system under the uncertainty of the target ship's movement in a complex multi-ship encounter situation. Based on the idea of model predictive control (MPC), this paper proposes an autonomous ship collision avoidance decision-making system (CADMS) suitable for the uncertainty of ship motion. The CADMS includes four modules: collision risk analysis module, control and execution module, ship trajectory prediction module and collision avoidance decision-making model. The proposed model can be implemented in the collision avoidance decision-making system for safe navigation or it can be included in the ship autonomous navigation process. The decision-making model is achieved from a risk identification-motion prediction-ship control-scheme implementation perspective, and the dynamic and uncertainty features of the ship action (i.e., alter course or change speed) are considered in the modelling process. The real-time rolling update of ship collision avoidance decisions is realised based on the time series rolling method. Four scenarios are designed to demonstrate the collision avoidance decision-making system's performance. The results show that the proposed collision avoidance decision-making system is a reasonable and effective system for collision avoidance, particularly in multi-ship encounter situations of target ships suddenly altering course or change speed.

TRFM-LS: Transformer-Based Deep Learning Method for Vessel Trajectory Prediction

In the context of the rapid development of deep learning theory, predicting future motion states based on time series sequence data of ship trajectories can significantly improve the safety of the traffic environment. Considering the spatiotemporal correlation of AIS data, a trajectory time window panning and smoothing filtering method is proposed for the abnormal values existing in the trajectory data. The application of this method can effectively deal with the jump values and outliers in the trajectory data, make the trajectory smooth and continuous, and ensure the temporal order and integrity of the trajectory data. In this paper, for the features of spatiotemporal data of trajectories, the LSTM structure is integrated on the basis of the deep learning Transformer algorithm framework, abbreviated as TRFM-LS. The LSTM module can learn the temporal features of spatiotemporal data in the process of computing the target sequence, while the self-attention mechanism in Transformer can solve the drawback of applying LSTM to capture the sequence information weakly at a distance. The advantage of complementarity of the fusion model in the training process of trajectory sequences with respect to the long-range dependence of temporal and spatial features is realized. Finally, in the comparative analysis section of the error metrics, by comparing with current state-of-the-art methods, the algorithm in this paper is shown to have higher accuracy in predicting time series trajectory data. The research in this paper provides an early warning information reference for autonomous navigation and autonomous collision avoidance of ships in practice.