This paper was published through Galoá and has a deposited DOI. To cite this paper, use one of the standards below:
In case you are one of the co-authors and want to register this paper in your Lattes, use the following code: doi > 10.17648/sobena-2024-195521
If you've NEVER registered a DOI in your Lattes, check our tutorial!To ensure the stability and proper functioning of offshore units, the equipment on board must operate fully and continuously. One of the ways to avoid failures or unexpected stops is by monitoring the vibration levels of these equipment. Vibration signals are collected through sensors, over a wide frequency range, depending on the equipment and the types of failures to be assessed. However, visual analysis of these vibration signals presents a challenge for the classification of possible failures, as these signals can present similar vibration spectra for different types of failures. One of the ways to automatically classify faults is through the use of artificial intelligence, where machine learning algorithms are trained using a set of statistical characteristics extracted from the data collected, both in the time and frequency domains. This work began with a failure database (COMFAULDA), obtained from a rotating machine simulator (rotor kit), where scenarios were simulated that represent, at different levels, normal operating conditions, shaft unbalance, shaft misalignment and a combination of unbalance and misalignment, to represent the most common types of failures in rotating machines. Data collection involved the use of a tachometer to monitor shaft rotation and accelerometers to acquire vibration data. From these vibration signals, 109 features were extracted in the time and frequency domains. The main objective of this study was to evaluate and select the number of most relevant features using the Kruskal-Wallis, mRMR (minimum redundancy, maximum relevance) and ReliefF techniques to be used as input for the classifier. After selecting the most relevant attributes, they were applied as input to the Artificial Neural Networks (ANN) and Support Vector Machines (SVM) classifiers with the purpose of identifying the minimum set of fault characteristics that lead to an accuracy above 95% in the classification of flaws present in COMFAULDA. The use of classification algorithms has proven to be highly effective in identifying combined faults leading to greater reliability in accurately identifying potential problems before they become more serious, which improves overall equipment performance and maintenance efficiency. Furthermore, the results also showed that it was possible to reduce the number of features needed to classify faults with both ANN and SVM, and that the Kruskal-Wallis, mRMR and ReliefF techniques can be used for this task.
With nearly 200,000 papers published, Galoá empowers scholars to share and discover cutting-edge research through our streamlined and accessible academic publishing platform.
Learn more about our products:
This proceedings is identified by a DOI , for use in citations or bibliographic references. Attention: this is not a DOI for the paper and as such cannot be used in Lattes to identify a particular work.
Check the link "How to cite" in the paper's page, to see how to properly cite the paper