الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 192 |  |  |
| | | from | 192 | | |
Abstract
The advent of Industry 4.0, synonymous with the Fourth Industrial Revolution, has propelled Prognostics and Health Management (PHM) as an essential facet within the domain of industrial big data and smart manufacturing. This study endeavors to demonstrate a proof-of-concept that elucidates the efficacy of employing machine learning methodologies for analyzing industrial facility data to forecast the condition of machinery.
Specifically, the research delves into a comprehensive case study centered on vibration monitoring. The primary goal is to anticipate maintenance requirements for the forced blower within a chemical plant by harnessing vibration data accumulated during the manufacturing process. To achieve this, the study employs a range of machine learning algorithms – Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Extreme Gradient Boosting (XGBoost), Multilayer Perceptron (MLP), and Random Forest (RF).
Validation of the methodology involves the utilization of evaluation metrics such as the Matthews Correlation Coefficient (MCC) and Receiver Operator characteristic Curve (ROC). The core aim here is to establish a tangible relationship between machine failures attributed to vibrations and the predictive capabilities of these machine-learning approaches.
The culmination of the study’s findings points toward the superiority of the Multilayer Perceptron algorithm among the selected methodologies. This algorithm notably outperforms others, showcasing an MCC of 0.850 along with a notably higher area under the ROC curve. These results affirm the efficacy and potential of utilizing machine learning, particularly the MLP approach, in prognosticating machinery health and predicting potential failures based on vibration analysis.