Search

In industrial settings, machinery components inevitably wear and degrade due to friction between moving parts. To address this, various maintenance strategies, including corrective, preventive, and predictive maintenance, are commonly employed. This paper focuses on predictive maintenance through vibration analysis, utilizing data-driven models. This study explores the application of unsupervised learning methods, particularly Convolutional Autoencoders (CAEs) and variational Autoencoders (VAEs), for anomaly detection (AD) in vibration signals. By transforming vibration signals into images using the Synchrosqueezing Transform (SST), this research leverages the strengths of convolutional neural networks (CNNs) in image processing, which have proven effective in AD, especially at the pixel level. The methodology involves training CAEs and VAEs on data from machinery in healthy condition and testing them on new data samples representing different levels of system degradation. The results indicate that models with spatial latent spaces outperform those with dense latent spaces in terms of reconstruction accuracy and AD capabilities. However, VAEs did not yield satisfactory results, likely because reconstruction-based metrics are not entirely useful for AD purposes in such models. This study also highlights the potential of ReLU residuals in enhancing the visibility of anomalies. The data used in this study are openly available. [ABSTRACT FROM AUTHOR]

This paper introduces a framework to perform operational modal analysis (OMA) for structural health monitoring (SHM) by presenting the development and validation of a low-power, solar-powered wireless sensor network (WSN) tailored for bridge structures. The system integrates accelerometers and temperature sensors for dynamic structural assessment, all interconnected through the energy-efficient message queuing telemetry transport (MQTT) messaging protocol. Moreover, it delves into the details of sensor selection, calibration, and the design considerations necessary to address the unique challenges associated with bridge structures. Special attention is given to the solar-powered aspect, allowing for extended deployment periods without the need for frequent maintenance or battery replacements. To validate the proposed system, a comprehensive field deployment was conducted on an actual bridge structure. The collected data were transmitted through MQTT messages and analyzed by means of OMA. Comparative studies with traditional wired systems underscore the advantages of the solar-powered wireless solution in terms of sustainability, scalability, and ease of deployment. Results from the validation phase demonstrate the system's capability to provide accurate and real-time data needed to assess the health state of the monitored asset. This paper concludes with insights into the practical implications of adopting such a solar-powered WSN, emphasizing its potential to revolutionize bridge health monitoring by offering a cost-effective and energy-efficient solution for long-term infrastructure resilience. [ABSTRACT FROM AUTHOR]

In the framework of direct Structural Health Monitoring approaches, this paper presents a low frequency range analysis performed on a steel truss bridge designed in 1946 and built up in Northern Italy. The bridge is instrumented with a permanent structural health monitoring system, that guarantees a continuous flux of data and information from a heterogeneous set of sensors. In this analysis, static and quasistatic bridge responses are considered. These experimental data are merged with the analogous outputs obtained from numerical simulations, performed on FE models representative of the actual bridge, with the purpose of identifying damage‐sensitive indexes. Simulations are performed considering the healthy structure and modelling different damage scenarios. The static analysis consists in the construction of a regressive model able to estimate sensor output responses starting from environmental measurements, with the primary purpose to reduce the effect of temperature over measured static response. This procedure allows to relate the damages simulated by means of a numerical model with the real data, so that the monitoring rules can effectively defined. [ABSTRACT FROM AUTHOR]