Risk Assessment of Ships in an Approaching Channel

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

MOST109-2410-H019-012

Translated Name/計畫中文名

船舶進出港航行即時風險評估

Project Coordinator/計畫主持人

Juan-Chen Huang

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Merchant Marine

Website

https://www.grb.gov.tw/search/planDetail?id=13537187

Year

2020

Start date/計畫起

01-08-2020

Expected Completion/計畫迄

31-07-2021

Bugetid/研究經費

488千元

ResearchField/研究領域

經濟學

本研究計畫目標是結合AIS數據資料與航道氣象數據資料庫，運用人工智慧演算法，開發智慧航道即時風險預警與監控系統，建置智慧化VTS系統。航管中心管制員依據船舶動態資料與環境即時資料，自動模擬船舶進出港或通過狹窄及繁忙水域時的動態過程，即時評估航道船舶進出港航行風險，發送航行安全警告通知，避免船舶發生擱淺或碰撞危機。提升VTS站台風險評估與有效監管能量，確保進出港航道、錨泊區與港區的安全。本計畫預計執行期程二年，第一年研究內容：依據船型、大小與氣象條件分類，設定資料擷取閘線，擷取船舶位置、航向角與船速等時序資料，進行航道資料統計分析與機率密度函數建構，利用Monte Carlo模擬法計算危險度函數，完成航道航行危險度模式建構方法開發。第二年研究內容為：基於AIS資料，結合神經網路和深度學習，提出二種適用於船舶軌跡預測的演算法模型。首先結合遺傳演算法與神經網路發展GA-BP淺層次船舶軌跡預測模型，加速學習過程收斂獲得全域神經網路模型。其次利用深度學習及時間序列的特性，建立迴圈神經網路-長短期記憶(RNN-LSTM)模型，並應用航跡預測模型在異常預警與航線規劃。The object of this project is to develop a real-time intelligent early warnling and monitoring system with the AIS data and the weather data of approaching channel for the intelligent Vessel Traffic Service (VTS) by using artificial intelligent algorithm. According to the ship's dynamic data and environmental real-time data, the VTS controller could predict the trajectories of ships in approaching channel and entrance, assessing the grounding or impact risks of ships. Moreover, the controller sends a warning notification immediately to avoid accidents if the system calculates the risk above a threshold. The real-time warning system proposed in this study can effectively improve the risk assessment and supervision capabilities of VTS, and ensure the safety of inbound and outbound ships.The project is expected to be implemented for two years. In the first year, we will statistically study the time series data such as ship position, heading angle, and ship speed which acquised from AIS data based on classification of ship types, sizes, and meteorological conditions. After the statistical model is completed, the Monte Carlo simulation method is used to calculate the risk function. With the risk function, a real-time intelligent early warnling and monitoring system for ships in approaching channel is implemented.In the second year, we will propose two algorithms combined with neural network and deep learning for ship trajectory prediction and based on AIS data. One is GA-BP algorithm, which combining back-propagation neural networks with genetic algorithms for ship trajectory prediction. GA-BP algorithm is expected to accelerate the learning process and is a global neural network model. The second one is the recurrent neural networkwith long short-term memory (RNN-LSTM) algorithm, which implemented by using the characteristics of deep learning and time series.The results of this research can be applied to the simulation and analysis of ship navigating in specific waters, to provide route planning and ship maneouvering recommendations, improve navigation safety and port traffic control.

Keyword(s)

AIS資料
VTS系統
船舶軌跡預測
Monte Carlo模擬法
GA-BP模型
RNN-LSTM模型
即時風險評估
AIS data
VTS system
ship trajectory prediction
Monte Carlo simulation method
GA-BP model
RNN-LSTM model
real-time risk assessment

DSpace CRIS

Risk Assessment of Ships in an Approaching Channel

Details

Description