Detection of Rupture Damage Degree in Laminated Rubber Bearings Using a Piezoelectric‐Based Active Sensing Method and Hybrid Machine Learning Algorithms

Structural Control and Health Monitoring, Journal Year: 2025, Volume and Issue: 2025(1)

Published: Jan. 1, 2025

Laminated rubber bearings may exhibit rupture damage due to factors such as temperature variations and seismic activity, which can reduce their isolation performance. Current detection methods, including human‐vision inspection computer‐vision inspection, have certain limitations in accurately assessing the degree of damage. This study attempts combine piezoelectric‐based active sensing method with a machine learning algorithm detect laminated bearings. A series varying degrees were fabricated, 1440 sets signals obtained through experiments using method. proposes hybrid that integrates one‐dimensional convolutional neural network (1DCNN), long short–term memory (LSTM) network, Bayesian optimization (BO) algorithm, extreme gradient boosting (XGB) algorithm. The involves 1DCNN LSTM algorithms extract deep features from wavelet packet energy spectra signals, then employing XGB optimized by BO construct prediction model. research results indicate proposed 1DCNN–LSTM–BO–XGB model achieved an accuracy value 98.6% on test set, outperforming 1DCNN–LSTM (91.7%), (88.9%), (25.0%), (90.3%), SVM (66.7%) algorithms. Therefore, combination shows promising application prospects detecting

Language: Английский

Reservoir Fluid Identification Based on Bayesian-Optimized SVM Model

Processes, Journal Year: 2025, Volume and Issue: 13(2), P. 369 - 369

Published: Jan. 28, 2025

Tight sandstone reservoirs are characterized by fine-grained rock particles, a high clay content, and complex interplay between the electrical properties gas content. These factors contribute to low-contrast reservoirs, where logging responses of water layers similar, resulting in traditional interpretation charts exhibiting low accuracy fluid-type classification. This inadequacy fails meet fluid identification needs study area’s severely restricts exploration development unconventional oil resources. To address this challenge, proposes method based on Bayesian-optimized Support Vector Machine (SVM) enhance efficiency reservoirs. Firstly, through sensitivity analysis responses, sensitive parameters such as natural gamma, compensated density, neutron, sonic logs selected input data for model. Subsequently, Bayesian optimization is employed automatically search optimal combination hyperparameters SVM Finally, an model established using optimized classify identify following four types: layers, gas–water dry layers. The proposed applied area, comparative experiments conducted with K-Nearest Neighbor (KNN), Random Forest (RF), AdaBoost models. classification performance each systematically evaluated metrics accuracy, recall, F1-score. experimental results indicate that outperforms other models identification, achieving average 91.41%. represents improvements 16.94%, 4.39%, 8.30% over KNN, RF, models, respectively. findings validate superiority area provide efficient feasible solution tight

Language: Английский

Research Status and Prospects of Intelligent Logging Lithology Identification

Measurement Science and Technology, Journal Year: 2024, Volume and Issue: 36(1), P. 012010 - 012010

Published: Dec. 10, 2024

Abstract With the increasing of petroleum exploration and development, accurate lithology identification is crucial. Machine learning (ML) plays a key role in logging identification. By introducing traditional methods, we review application ML from perspectives bibliometrics classification this paper. The applications supervised learning, semi-supervised unsupervised ensemble deep algorithms are introduced detail. Multiple have achieved remarkable results different scenarios. For example, support vector machine, random forest, eXtreme gradient boosting, convolutional neural network perform well obtain relatively high accuracy. However, for also faces challenges such as data quality, imbalance, model generalization, interpretability. Future research should focus on algorithm optimization innovation, improvements quality quantity, multidisciplinary integration practical to enhance accuracy reliability These findings provide strong oil gas development.

Language: Английский