Total views : 132
Design and Implementation of Marine Elevator Safety Monitoring System based on Machine Learning
Objectives: Marine elevators require the mechanism of safety management due to the difficulty of their maintenance in voyage. It is necessary to monitor the status of elevators using sensors and to provide the maintenance information locally and remotely. Methods/Statistical Analysis: We designed and implemented a safety elevator monitoring system based on the NMEA 2000 network, which implements maintenance prediction in the gateway using big data on the server. Findings: We conducted supervised learning using labeled data, which are results of event messages from In-network processing module of logging gateway. The accuracy of load and platform tilt based slope prediction model is 0.99 or above, but the accuracy of roll and pitch based slope prediction model is below 0.94. Therefore, it cannot be adapted to the logging gateway prediction model. Features for prediction models have a very important role in accuracy of results during experimenting and we can use highly accurate prediction models in logging gateway to analyze the sensor data. Improvements/Applications: In the decision accuracy of operating condition, 0.95 or above accuracy is obtained using an operating time feature having the ±1 second margin. But for more high accuracy, we need experts’ analysis to modify the prediction model. This system can automate the decision of elevator problems using sensor monitoring and diagnosis prediction model.
Diagnosis Prediction, Elevator Safety Maintenance, Logging Gateway, Marine Elevator, Sensor Monitoring
- Jayavel K, Nagarajan V. Survey of Migration, Integration and Interconnection Techniques of Data Centric Networks to Internet- Towards Internet of Things (IoT), Indian Journal of Science and Technology. 2016 February; 9(8):1–7.
- Nick Heath. Keeping the world’s elevators running smoothly with machine learning and IoT. Posted in March 2, 2015: Available from: http://www.zdnet.com/article.
- NMEA, NMEA 2000 - Appendices A & B - Parameter Groups (PGNs) NMEA Network Messages, VERSION 2.000, January 2013.
- Aggarwal, Charu C. (Ed.), Managing and Mining Sensor Data, Springer 2016.
- Roman Kern. Feature Engineering. http://kti.tugraz.at/ staff/denis/- courses/kddm1/featureengineering.pdf. Date accessed: 29/10/2015.
- Maheshwari V, Prasanna M. Generation of Test Case using Automation in Software Systems – A Review, Indian Journal of Science and Technology. 2015 Dec; 8(35):1–8.
- DNV(1987), Rules for Certification of Lifts in Ships, Mobile Offshore Units and Offshore Installations, 1987.
- DMLC(Distributed Machine Learning Community) XGBoost Parameters. 2016. https://github.com/dmlc/xgboost/ blob/master/doc/parameter.md.
- Cynthia Deb, M. Ramesh Nachiappan, M. Elangovan, V.Sugumaran. Fault Diagnosis of a Single Point Cutting Tool using Statistical Features by Random Forest Classifier. Indian Journal of Science and Technology. 2016; 9(33):1–8.
- Preeti Aggarwal and Deepak Dahiya. Contribution of Four Class Labeled Attributes of Kdd Dataset on Detection and False Alarm Rate for Intrusion Detection System. Indian Journal of Science and Technology. 2016; 9(5):1–8.
- Wikipedia, Cohen’s kappa. 2016a. https://en.wikipedia.org/ wiki/Cohen%27s_kappa.
- There are currently no refbacks.
This work is licensed under a Creative Commons Attribution 3.0 License.