Font Size: a A A

Research And Implementation Of Key Technologies Of Ship Intelligent Monitoring And Early Warning Platform

Posted on:2021-03-30Degree:MasterType:Thesis
Country:ChinaCandidate:P WangFull Text:PDF
GTID:2392330611452092Subject:Engineering, Electronics and Communication Engineering
Abstract/Summary:PDF Full Text Request
Ship is a huge "man-ship-ocean" complex system with multiple factors coupling.The monitoring and early warning of it are particularly important for such a large and complex system.The occurrence of a large number of ship accidents every year and the rise of the fourth-generation industrial revolution prompts the urgent need to design and develop a modern ship intelligent monitoring and early warning platform: In order to complete the distributed networking of ship sensors and real-time acquisition of ship status data,communication technology and electronic technology are used through the deployment of sensors in key parts of the ship,and apply artificial intelligence technology to learning the collected ship data,so that the crew can find and control the ship to ensure the health of the safety of ship and avoid the occurrence of major marine accidents.This paper designs and develops a ship intelligent monitoring and early warning platform based on artificial intelligence.First,according to the data requirements of ship monitoring and early warning,a sensor-node-controller was designed for unified management,unified access,and unified networking is designed to complete the collection of ship status data.Then,a comprehensive outlier detection model(CODM)based on statistical rules,unsupervised learning and deep learning is proposed to establish a three-dimensional of ship anomaly detection,offline learning and online warning;Finally,a set of ship monitoring and early warning software was developed based on the designed hardware platform and algorithm platform,which realized the whole process of ship data collection,storage,monitoring and early warning.The main innovations and work of this article are as follows:(1)In terms of hardware design: the sensor-node-controller developed in this paper integrates multiple communication methods and communication protocols on a single platform,and the sensor network built whth the controller are tailorable,scalable and building blocks.First of all,we investigate and analyze the sensors on the ship.Then,according to the characteristics of ship sensors,we design and develop a sensor-node-controller to acquisition of ship data.Finally,we set up a ship sensor network to collect the real-time data of the ship,Including ship speed,heading,attitude,position,temperature,pressure and so on.(2)In terms of algorithm design: this paper proposes a ship state early warning method based on a comprehensive anomaly detection model,we integrate multiple time series anomaly detection algorithms to improve the reliability and robustness of early warning.First,the feature extraction method of sensor time series is studied,and the residual sequence extraction method based on Long Short-Term Memory(LSTM)is proposed.We study a comprehensive anomaly detection method based on statistical rules,unsupervised learning and deep learning to realize anomaly decision,the platform will send an early warning message to prompt the crew to inspect the ship and take corresponding control measures when the decision result of CODM model is abnormal.(3)In terms of software design: This paper designs and develops a set of monitoring and early warning software based on C sharp.This software can realize the whole process of ship state data collection,storage,monitoring and early warning.We have completed the real-time monitoring of the ship's state,and realized the monitoring and early warning modules of ship navigation and ship health.The experiment shows that the platform can assist the crew very well and it can complete the real-time monitoring and early warning of ship status to avoid the occurrence of major shipwreck accidents.
Keywords/Search Tags:ship networking, multi-sensor networking, intelligent monitoring and early warning, unsupervised learning, deep learning
PDF Full Text Request
Related items