基于物联网架构的智能航海避碰算法研究

发布时间：2018-05-14 07:36

本文选题：物联网 + 智能航海避碰　；参考：《天津理工大学》2017年硕士论文

【摘要】：随着“一带一路”战略的提出,“21世纪海上丝绸之路”构想将中国海运事业提升至一个新高度。海运量增长的同时也给安全航海带来前了所未有的挑战。目前,航海安全主要依靠值班驾驶员在雷达、自动识别系统(AIS)、自动雷达标绘仪(ARPA)等辅助设备协助下对碰撞危险进行判断。但船-船相碰事故有增无减,带来极大的财产损失和人员伤亡,如何更有效地预防海上船舶碰撞事故的发生成为亟待解决的问题。为提高航海安全性,各国学者对智能避碰决策做了大量工作并取得一定成果。而随着物联网技术在智能交通、智能家居、健康监测等多个领域的逐步渗入,整个社会变得更加智能化。本文通过深入研究物联网和智能算法的相关理论,将其关键技术综合应用到智能航海避碰中。提出了基于物联网架构的智能航海避碰系统模型,设计了一种可以实时预测船舶碰撞风险度的智能航海避碰算法,该算法可提前“预知”船舶在下一时刻抵达位置处的碰撞风险程度并在危险来临前推送避碰优先序列。本文所做的工作如下:(1)提出了基于物联网架构的智能航海避碰系统模型。该模型包括感知层、传输层和应用层。其中,感知层以性价比更高的智能传感器模块为主体对各类航海信息进行实时采集;传输层以物联网中间件及Zigbee协议为基础进行信息传输。最后由应用层的智能避碰算法分析、预测船舶碰撞风险并作出避碰优先序列决策。(2)通过深入研究国内外智能避碰算法,基于层次分析法(AHP)和BP人工神经网络算法设计了一种改进的智能航海避碰算法。该算法加入了环境因子,使整个算法可根据实际环境的改变而实时调整策略,具有实际应用价值。首先,本文基于AHP确定影响安全航海的各环境因素所占的权重,将采集到的各影响因素值与其权重相结合,动态确定环境因子;根据感知层采集到的实时数据与环境因子综合确定最佳预测时间T。其次,运用BP神经网络算法学习船舶的航海习惯;基于训练完成的BP神经网络预测T时刻后船舶可能抵达的位置坐标P。最后,根据航海学知识建立模型,判断P处的碰撞风险度;根据不同船舶与本船之间碰撞的风险度大小建立避碰优先列表,供值班驾驶员参考。(3)本文基于MATLAB平台对改进的智能航海避碰算法所有步骤进行建模仿真,以渤海海域内渔船、货船等实际船舶的历史航行记录数据作为测试用例进行实验,评估验证本文提出的改进算法的可用性和有效性。
[Abstract]:With the development of "Belt and Road" strategy, the concept of "21st Century Maritime Silk Road" will elevate China's maritime transport industry to a new height. The increase in sea traffic also poses a challenge to safe navigation. At present, navigational safety mainly depends on the assistant equipment such as radar, automatic identification system (AISN), automatic radar plotter (ARPAA) and so on, to judge collision risk. However, the accidents of ship-ship collision are increasing, which brings great loss of property and casualties. How to prevent the collision of ships at sea more effectively becomes an urgent problem to be solved. In order to improve the safety of navigation, scholars from various countries have done a lot of work on intelligent collision avoidance decision and achieved certain results. With the gradual infiltration of Internet of things technology in intelligent transportation, smart home, health monitoring and other fields, the whole society has become more intelligent. In this paper, the relevant theories of Internet of things and intelligent algorithm are deeply studied, and its key technologies are applied to intelligent navigation collision avoidance. This paper presents an intelligent navigation collision avoidance system model based on the Internet of things, and designs an intelligent navigation collision avoidance algorithm which can predict ship collision risk in real time. The algorithm can predict the collision risk of the ship at the next moment and push the collision avoidance priority sequence before the danger comes. The work of this paper is as follows: 1) A model of intelligent navigation collision avoidance system based on the Internet of things is proposed. The model includes perceptual layer, transport layer and application layer. Among them, the sensing layer takes the intelligent sensor module with higher performance and price ratio as the main body to collect all kinds of navigation information in real time; the transmission layer is based on the middleware of the Internet of things and the Zigbee protocol for information transmission. Finally, by analyzing the intelligent collision avoidance algorithm of application layer, forecasting ship collision risk and making collision avoidance priority sequence decision. Based on the analytic hierarchy process (AHP) and BP artificial neural network (Ann), an improved intelligent navigation collision avoidance algorithm is designed. The environment factor is added to the algorithm, which makes the whole algorithm can adjust the strategy in real time according to the change of the actual environment, so it has practical application value. First of all, based on AHP, this paper determines the weight of the environmental factors which affect the safe navigation, and combines the collected values of the factors with their weights to determine the environmental factors dynamically. According to the real-time data collected from the perception layer and the environmental factors, the best prediction time T is determined. Secondly, the BP neural network algorithm is used to study the navigation habits of ships, and based on the BP neural network completed by the training, the coordinates of the ship's possible arrival position after T time can be predicted. Finally, according to the navigational knowledge, the collision risk degree at P is judged, and the collision avoidance priority list is established according to the collision risk degree between different ships and their own ships. This paper models and simulates all steps of the improved intelligent navigation collision avoidance algorithm based on MATLAB platform, and takes the historical navigation record data of fishing vessels and freighters in the Bohai Sea as test cases. The evaluation verifies the availability and effectiveness of the proposed improved algorithm.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.44;TN929.5

【相似文献】