Financial support for this research was provided in part by the National Science Foundation (NSF) under NSF grants CMS 03-01140 and CMS 06-00433 (Dr. S. C. Liu, Program Manager). The first author was supported by a Vodafone-U.S. Foundation Graduate Fellowship. This support is gratefully acknowledged.

http://www.ideals.illinois.edu/bitstream/handle/2142/3521/NSEL.Report.001.pdf?sequence=4

Structural Health Monitoring Using Smart Sensors

Wireless smart sensors enable new approaches to improve structural health monitoring (SHM) practices through the use of distributed data processing. Such an approach is scalable to the large number of sensor nodes required for high-fidelity modal analysis and damage detection. While much of the technology associated with smart sensors has been available for nearly a decade, there have been limited numbers of full- scale implementations due to the lack of critical hardware and software elements. This research develops a flexible wireless smart sensor framework for full-scale, autonomous SHM that integrates the necessary software and hardware while addressing key implementation requirements. The Imote2 smart sensor platform is employed, providing the computation and communication resources that support demanding sensor network applications such as SHM of civil infrastructure. A multi-metric Imote2 sensor board with onboard signal processing specifically designed for SHM applications has been designed and validated. The framework software is based on a service-oriented architecture that is modular, reusable and extensible, thus allowing engineers to more readily realize the potential of smart sensor technology. Flexible network management software combines a sleep/wake cycle for enhanced power efficiency with threshold detection for triggering network wide operations such as synchronized sensing or decentralized modal analysis. The framework developed in this research has been validated on a full-scale a cable-stayed bridge in South Korea.

Flexible smart sensor framework for autonomous structural health monitoring

Data-driven methods in structural health monitoring (SHM) is gaining popularity due to recent technological advancements in sensors, as well as high-speed internet and cloud-based computation. Since the introduction of deep learning (DL) in civil engineering, particularly in SHM, this emerging and promising tool has attracted significant attention among researchers. The main goal of this paper is to review the latest publications in SHM using emerging DL-based methods and provide readers with an overall understanding of various SHM applications. After a brief introduction, an overview of various DL methods (e.g., deep neural networks, transfer learning, etc.) is presented. The procedure and application of vibration-based, vision-based monitoring, along with some of the recent technologies used for SHM, such as sensors, unmanned aerial vehicles (UAVs), etc. are discussed. The review concludes with prospects and potential limitations of DL-based methods in SHM applications.

Data-Driven Structural Health Monitoring and Damage Detection through Deep Learning: State-of-the-Art Review.

https://ir.lib.uwo.ca/cgi/viewcontent.cgi?article=1193&context=electricalpub

A systematic review of convolutional neural network-based structural condition assessment techniques

In recent years, deep learning techniques have outperformed traditional models in many machine learning tasks. Deep neural networks have successfully been applied to address time series forecasting problems, which is a very important topic in data mining. They have proved to be an effective solution given their capacity to automatically learn the temporal dependencies present in time series. However, selecting the most convenient type of deep neural network and its parametrization is a complex task that requires considerable expertise. Therefore, there is a need for deeper studies on the suitability of all existing architectures for different forecasting tasks. In this work, we face two main challenges: a comprehensive review of the latest works using deep learning for time series forecasting and an experimental study comparing the performance of the most popular architectures. The comparison involves a thorough analysis of seven types of deep learning models in terms of accuracy and efficiency. We evaluate the rankings and distribution of results obtained with the proposed models under many different architecture configurations and training hyperparameters. The datasets used comprise more than 50,000 time series divided into 12 different forecasting problems. By training more than 38,000 models on these data, we provide the most extensive deep learning study for time series forecasting. Among all studied models, the results show that long short-term memory (LSTM) and convolutional networks (CNN) are the best alternatives, with LSTMs obtaining the most accurate forecasts. CNNs achieve comparable performance with less variability of results under different parameter configurations, while also being more efficient.

An Experimental Review on Deep Learning Architectures for Time Series Forecasting.

Deep learning has ushered in many breakthroughs in vision‐based detection via convolutional neural networks (CNNs), but the vibration‐based structural damage detection by CNN remains being...

Vibration‐based structural state identification by a 1‐dimensional convolutional neural network

Smart sensors densely distributed over structures can provide rich information for structural monitoring using their onboard wireless communication and computational capabilities However, issues such as time synchronization error, data loss, and dealing with large amounts of harvested data have limited the implementation of full-fledged systems Limited network resources (eg battery power, storage space, bandwidth, etc) make these issues quite challenging This paper first investigates the effects of time synchronization error and data loss, aiming to clarify requirements on synchronization accuracy and communication reliability in SHM applications Coordinated computing is then examined as a way to manage large amounts of data

/pdf/issues-in-structural-health-monitoring-employing-smart-1sk5acjadw.pdf

Issues in structural health monitoring employing smart sensors

This paper deals with the detection of damage in a lightly reinforced concrete beam using dynamic measurements. An algorithm based on changes in the power spectral density (PSD) is presented. The algorithm is used to detect damage, predict the location and assess the extent of damage in the reinforced concrete beam. The method is based on the measured data on the beam after introducing some damage at salient locations. A tuneable piezoelectric actuator was used to produce additional peaks in the dynamic response of the system in the frequency domain. These additional peaks were adjusted over the frequency band by tuning the actuator. It was noticed from the results, that the PSD method detected the damage, determine the location and could monitor the increase in damage in the beam

Damage identification in a lightly reinforced concrete beam based on changes in the power spectral density

Identification of a human walking force model based on dynamic monitoring data from pedestrian bridges

The Bridge Engineering Laboratory in Kitami Institute of Technology, Japan has introduced a number of different damage identification techniques to detect structural damage and identify its location utilizing piezoelectric actuators as a localized excitation source. Several spectral functions, such as cross spectral density, power spectral density, phase angle and transfer function estimate, were used to estimate the dynamic response of the structure. Each function’s magnitude, measured in a specified frequency range, is used in the damage identification methods. The change of the spectral function magnitude between the baseline state and the current state is then used to identify the location of possible damage in the structure. It is then necessary to determine which spectral function is best able to estimate the dynamic response and which algorithm is best able to identify the damage. The first part of this paper compares the performance of different spectral functions when their magnitude is used in one damage identification algorithm using experimental data from a railway steel bridge. The second part of this paper compares the performance of different damage identification algorithms using the same data.

Yasunori Miyamori

Papers

Vibration‐based structural state identification by a 1‐dimensional convolutional neural network

Issues in structural health monitoring employing smart sensors

Damage identification in a lightly reinforced concrete beam based on changes in the power spectral density

Identification of a human walking force model based on dynamic monitoring data from pedestrian bridges

Assessment of vibration-based damage identification techniques using localized excitation source