Article
Computer Science, Information Systems
Frank Billy Djupkep Dizeu, Michel Picard, Marc-Antoine Drouin, Guillaume Gagne
Summary: This paper proposes a method for drone detection using propeller rotation speed as a key parameter, extracting drone features through discrete Fourier transform to accurately distinguish drones from other flying entities. Experiment results from a consumer-grade camera at 240Hz frame rate demonstrate the effectiveness and reliability of this method.
Article
Computer Science, Artificial Intelligence
Xiao Lang, Da Wu, Wengang Mao
Summary: This paper proposes a novel physics-informed machine learning method to build a grey-box model for predicting ship speed. The method combines physics-informed neural networks to model expected ship speed in calm water with the eXtreme Gradient Boosting algorithm to estimate speed reduction under actual weather conditions. Experimental results show that the proposed grey-box model can significantly improve speed prediction accuracy and perform well even with limited data.
EXPERT SYSTEMS WITH APPLICATIONS
(2024)
Article
Engineering, Marine
Xinqiang Chen, Chenxin Wei, Guiliang Zhou, Huafeng Wu, Zhongyu Wang, Salvatore Antonio Biancardo
Summary: Automatic Identification System (AIS) data-supported ship trajectory analysis plays a crucial role in helping maritime regulations and practitioners make informed traffic controlling and management decisions. This study proposes a novel ship trajectory exploitation and prediction framework using the bidirectional long short-term memory (Bi-LSTM) model, which shows satisfactory prediction performance according to evaluation metrics.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Engineering, Marine
Sara El Mekkaoui, Loubna Benabbou, Stephane Caron, Abdelaziz Berrado
Summary: Improving maritime operations planning and scheduling is crucial for enhancing the performance and competitiveness of the sector. Accurate ship speed estimation is important for efficient maritime traffic management. This study proposes a data-driven solution using deep learning sequence methods and historical ship trip data, which outperforms the baseline ship speed rates. The findings suggest that deep learning models combined with maritime data can effectively estimate ship speed and improve operational efficiency, navigation safety, and ship emissions estimation and monitoring.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Marine
Ameen M. Bassam, Alexander B. Phillips, Stephen R. Turnock, Philip A. Wilson
Summary: This study investigates a data-driven machine learning approach for ship speed prediction through regression, utilizing the ship operational dataset of the M/S Smyril ferry. The results demonstrate that the proposed approach can accurately predict ship speed under real operational conditions and help optimize ship operational parameters.
Article
Engineering, Civil
Yongjun Chen, Ming Huang, Kaixuan Song, Tengfei Wang
Summary: Accurately predicting short-term congestions in ship traffic flow is crucial for water traffic safety and intelligent shipping. A method using the whale optimization algorithm and extreme learning machine is proposed, considering external environmental uncertainty and complexity of ships navigating in traffic-intensive waters. The method accurately predicts ship traffic congestion levels, with a prediction accuracy of 76.04% for ship traffic congestion. The effectiveness of the method is verified using ship traffic flow data of the Yangtze River estuary.
JOURNAL OF ADVANCED TRANSPORTATION
(2023)
Article
Engineering, Marine
Dangli Wang, Yangran Meng, Shuzhe Chen, Cheng Xie, Zhao Liu
Summary: Accurate vessel traffic flow prediction is crucial for maritime traffic guidance and control. A new hybrid model, DWT-Prophet, combining discrete wavelet decomposition and Prophet framework, is proposed for vessel traffic flow prediction. Experimental results show that DWT-Prophet outperforms other traditional methods, improving the practicality of the forecasting approach.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2021)
Article
Engineering, Civil
Maohan Liang, Ryan Wen Liu, Yang Zhan, Huanhuan Li, Fenghua Zhu, Fei-Yue Wang
Summary: Accurate and robust vessel traffic flow prediction is crucial in maritime intelligent transportation systems. To achieve this, a graph-driven neural network is proposed, which utilizes the maritime traffic network and extracts feature points from vessel positioning data. The method demonstrates superior performance in accuracy and robustness.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Marine
Yong Li, Zhaoxuan Li, Qiang Mei, Peng Wang, Wenlong Hu, Zhishan Wang, Wenxin Xie, Yang Yang, Yuhaoran Chen
Summary: The intelligent maritime transportation system plays a crucial role in port management due to advancements in artificial intelligence and big data technology. This paper proposes a data-driven approach to construct a multi-port network and designs a spatiotemporal graph neural network model to capture traffic variation patterns. Compared to existing methods, this model considers the network characteristics of the overall port and fills the research gap in multi-port scenarios.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Chemistry, Analytical
Duanyang Liu, Xinbo Xu, Wei Xu, Bingqian Zhu
Summary: This study introduces a novel graph convolutional network model FSTGCN to improve the accuracy of traffic speed prediction by incorporating historical traffic flow data and designing a dynamic adjacency matrix, achieving better forecasting results than existing methods.
Article
Engineering, Marine
Tian Xu, Qingnian Zhang
Summary: In this study, a Gated Recurrent Unit (GRU) of a Recurrent Neural Network (RNN) was used to analyze the changing characteristics of ship traffic flow in complex waters. A spatiotemporal dependence feature matrix was constructed to predict ship traffic flow, and the model was applied to the wind farms water area in Yancheng city, Jiangsu Province. The GRU method based on spatiotemporal dependence showed higher accuracy and reliability compared to traditional models.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2022)
Article
Engineering, Marine
Yuhan Guo, Yiyang Wang, Jiaqi Zhou, Jun Wang
Summary: Data-driven methods are widely used in short-term vessel speed prediction in ocean engineering. However, the requirement for large amounts of training data and the distributed inconsistency among datasets make it difficult to achieve generalization or superior performance. We propose a domain-adapted feature transfer (DAFT) method to design domain-invariant features and narrow the distribution differences between datasets in order to improve the effectiveness of data-driven models.
Article
Engineering, Marine
Ameen M. Bassam, Alexander B. Phillips, Stephen R. Turnock, Philip A. Wilson
Summary: Ship speed is a crucial parameter that impacts ship design, energy efficiency, and safety. Data-driven methodologies and Artificial Neural Network (ANN) techniques have gained significant attention for ship speed prediction due to their efficiency and accuracy. This study explores ANN models of various sizes and architectures to find the optimal network parameters for ship speed prediction. The results demonstrate that the proposed ANN model accurately predicts ship speed with an error of less than 1 knot in real operational conditions, providing valuable insights for voyage planning and execution optimization.
Article
Engineering, Multidisciplinary
Liang Chen, Jingsen Qi, Jin Shi
Summary: This paper presents an analysis and application method of ship traffic flow considering navigation rules in narrowing channel. It aims to provide a reasonable judgment for decision makers to improve navigation efficiency and ensure navigation safety. The relationship model between narrowing channel and ships is analyzed, and navigation rules are formulated for different positions of the channel. The mutual influence of ships at the narrowed position is also analyzed. A traffic flow model of narrowing channel with navigation safety distance is proposed using cellular automata algorithm, and its correctness is verified with measured data. The experimental results demonstrate that the method can effectively capture the characteristics of actual ship traffic flow in narrowing channels.
ALEXANDRIA ENGINEERING JOURNAL
(2023)
Article
Computer Science, Information Systems
Xu Luo, Fumin Zou, Sijie Luo, Feng Guo
Summary: Accurate modeling of travel speeds is crucial for optimizing roadway management. A study proposes an innovative framework that integrates machine learning prediction of service area dwell times into travel speed calculation to recover normal driving behavior and assess traffic conditions more effectively.
Article
Physics, Applied
Jiansen Zhao, Zhen Sun, Yuxiang Ren, Lu Song, Shengzheng Wang, Wei Liu, Zhe Yu, Yuhan Wei
JOURNAL OF PHYSICS D-APPLIED PHYSICS
(2019)
Article
Engineering, Electrical & Electronic
Xinqiang Chen, Shubo Wu, Chaojian Shi, Yanguo Huang, Yongsheng Yang, Ruimin Ke, Jiansen Zhao
IEEE SENSORS JOURNAL
(2020)
Article
Engineering, Marine
Shunan Hu, Shenpeng Tian, Jiansen Zhao, Ruiqi Shen
Summary: To ensure the safe navigation and real-time collision avoidance of unmanned surface vessels (USVs), this study conducts global and local path planning in a variable dynamic environment. The local path planning takes into account the USV motion characteristics and the requirements of COLREGs. The proposed local path planning algorithm is verified through simulation experiments with unknown obstacles, proving its effectiveness in ensuring collision avoidance decisions based on COLREGs when the USV encounters a ship.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Marine
Jiansen Zhao, Fengchuan Song, Guobao Gong, Shengzheng Wang
Summary: To accurately and in real-time monitor shorelines, an improved shoreline detection method is proposed, combining water surface area segmentation and edge detection. An improved U-Net network is introduced for water segmentation, using ResNet-34 as the backbone and a concise decoder integrated attention mechanism to enhance processing speed and accuracy. Transfer learning is also employed to improve training efficiency and address data insufficiency. Experimental results show that the network achieves 97.05% MIoU and 40 FPS with the fewest parameters and effectively detects shorelines in real time in various environments.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Chemistry, Analytical
Shunan Hu, Haiyan Duan, Jiansen Zhao, Hailiang Zhao
Summary: This study proposes a machine vision-based method for automatic extraction and evaluation of rust on navigation buoys. The method integrates image segmentation and processing, and analyzes the color transformation law to extract the rusted parts of the buoys for evaluation. Experimental results demonstrate high segmentation precision and accuracy, surpassing the performance of direct image processing-based rust evaluation algorithms.
Article
Engineering, Marine
Zhenzhen Zhou, Jiansen Zhao, Xinqiang Chen, Yanjun Chen
Summary: Obtaining ship navigation information from maritime videos can improve supervision efficiency and safety warnings. This study proposes a deep learning-based framework for ship speed extraction under the haze environment, which includes haze removal, ship detection and tracking, and speed estimation. Experimental results show that the proposed method effectively reduces haze interference and enhances image quality while extracting ship speed.
JOURNAL OF MARINE SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Information Systems
Qiang Luo, Jie Yuan, Xinqiang Chen, Junheng Yang, Wenhui Zhang, Jiansen Zhao
Article
Computer Science, Information Systems
Xinqiang Chen, Xueqian Xu, Yongsheng Yang, Huafeng Wu, Jinjun Tang, Jiansen Zhao