نوع مقاله : پژوهشی
تازه های تحقیق
عنوان مقاله English
نویسندگان English
In this study, a data-driven framework based on vibration signal analysis and a two-stage multilayer perceptron is proposed for bearing RUL prediction. First, the vibration signals are segmented into equal time intervals, and representative statistical features are extracted from each segment. Subsequently, a new feature-based time-series dataset is constructed by arranging the extracted statistical features in chronological order. To eliminate short-term fluctuations and capture the long-term degradation trend, these feature series are decomposed using Empirical Mode Decomposition, and the residual components are employed as health indicators. The main contribution of this study is a two-stage prediction framework in which the residuals extracted from horizontal vibration signals are first fed into an MLP to generate an initial RUL estimate. The predicted RUL is then used as an auxiliary feature together with the residuals extracted from vertical vibration signals and provided as inputs to a second MLP to produce the final prediction. By exploiting the complementary information contained in the two vibration directions, the proposed framework enables more accurate degradation modeling and improves prediction accuracy without utilizing the true RUL values or any future information during the prediction process. The proposed framework is evaluated on the benchmark PRONOSTIA bearing dataset, and the experimental results demonstrate its effectiveness and promising performance in predicting the remaining useful life of bearings.
کلیدواژهها English
Copyright © Ali Najmi, Rasoul Shafaei
License
This article is released under the Creative Commons Attribution (CC BY 4.0) license. Anyone is free to copy, share, translate, and adapt this article for any purpose, whether commercial or non-commercial, as long as proper citation is given to the authors and original publication.
