An official website of the United States government
Structures experience large vibrations and stress variations during their life cycles. This causes reduction in their load-carrying capacity which is the main design criteria for many structures. Therefore, it is important to accurately establish the performance of structures after construction that often needs full-field strain or stress measurements. Many traditional inspection methods collect strain measurements by using wired strain gauges. These strain gauges carry a high installation cost and have high power demand. In contrast, this paper introduces a new methodology to replace this high cost with utilizing inexpensive data coming from wireless sensor networks. The study proposes to collect acceleration responses coming from a structure and give them as an input to deep learning framework to estimate the stress or strain responses. The obtained stress or strain time series then can be used in many applications to better understand the conditions of the structures. In this paper, designed deep learning architecture consists of multi-layer neural networks and Long Short-Term Memory (LSTM). The network achieves to learn the relationship between input and output by exploiting the temporal dependencies of them. In the evaluation of the method, a three-story steel building is simulated by using various dynamic wind and earthquake loading scenarios. The acceleration time histories under these loading cases are utilized to predict the stress time series. The learned architecture is tested on acceleration time series that the structure has never experienced.
more »
« less
Structural sensing with deep learning: Strain estimation from acceleration data for fatigue assessment
Abstract Many of the civil structures experience significant vibrations and repeated stress cycles during their life span. These conditions are the bases for fatigue analysis to accurately establish the remaining fatigue life of the structures that ideally requires a full‐field strain assessment of the structures over years of data collection. Traditional inspection methods collect strain measurements by using strain gauges for a short time span and extrapolate the measurements in time; nevertheless, large‐scale deployment of strain gauges is expensive and laborious as more spatial information is desired. This paper introduces a deep learning‐based approach to replace this high cost by employing inexpensive data coming from acceleration sensors. The proposed approach utilizes collected acceleration responses as inputs to a multistage deep neural network based on long short‐term memory and fully connected layers to estimate the strain responses. The memory requirement of training long acceleration sequences is reduced by proposing a novel training strategy. In the evaluation of the method, a laboratory‐scale horizontally curved girder subjected to various loading scenarios is tested.
more »
« less
- PAR ID:
- 10155268
- Publisher / Repository:
- Elsevier
- Date Published:
- Journal Name:
- Computer-Aided Civil and Infrastructure Engineering
- Volume:
- 35
- Issue:
- 12
- ISSN:
- 1093-9687
- Format(s):
- Medium: X Size: p. 1349-1364
- Size(s):
- p. 1349-1364
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
High cycle fatigue is a major cause of cracking in steel structures subjected to cyclic loading. It can result in substantial financial losses and structural failures compromising the safety of users. Uniaxial methods are in many cases insufficient for large in-service structures with complex geometry and connections subjected to multiaxial non-proportional loadings. A new method for fatigue life prediction for complex structures is presented using the critical plane method and the Kalman filter. The applicability of the methodology proposed is demonstrated and evaluated in a roller coaster support structure. Strain rosettes and accelerometers were installed on a support bracket near weld lines to measure responses. A substructure model is defined and used to estimate response prediction in the weld of the support bracket. The estimation of the input and the state estimation is performed using the augmented Kalman filter method, based on the response measurements and the substructured model. This new methodology is anticipated to be used for real-time fatigue prognosis of highway bridges.more » « less
-
null (Ed.)Fatigue-induced damage is one of the most common types of damage experienced by civil engineering structures subjected to cyclic loading such as bridges and rollercoasters. A framework for the analysis of multiaxial fatigue damage using strain rosettes installed on welded connections is proposed. The applicability of this methodology is shown using strain measurements collected in a welded gussetless truss connection of a vertical-lift bridge. Commonly used uniaxial fatigue analysis methods are insufficient in complex structures that experience variable amplitude, multiaxial loading, and non-proportional loading. Strain data with these characteristics are used for the estimation of the number of multiaxial stress reversals induced by in service loads and the number of associated cycles using the rain-flow method. Methods proposed for uniaxial loading and multiaxial non-proportional loading are compared. Results show that non-proportional loading and the accuracy of the critical plane estimation can cause a significant decrease in the estimates of remaining fatigue life. The methodology proposed is anticipated to be used for real-time fatigue prognosis aiming to address critical needs related to maintenance procedures of complex structures, visual inspection techniques and evaluation tools for infrastructure networks.more » « less
-
Abstract Characterizing materials under shock loading has been of interest in fields such as protective material development, biomechanics to study the injury mechanics and high-speed aerodynamic structures. However, shock loading of material is a very short duration phenomenon and it is extremely challenging to develop sensors for dynamic measurements under such loading conditions. Optical fiber sensors present the possibilities to allow high resolution measurement of displacement in such high strain rate loading conditions. This work studies the possibility of using a fiber-optic loop sensor (FOLS) based on the principle of power losses from the curved section for dynamic measurements under shock loading conditions. The displacement results obtained from the optical sensors are compared with the traditional strain gauge and digital image correlation (DIC) measurements. The result obtained by the FOLS closely matched the sensitivity and precision of the strain gauges and had higher precision than that of DIC.more » « less
-
In this work, we develop a two time-scale deep learning approach for beamforming and phase shift (BF-PS) design in time-varying RIS-aided networks. In contrast to most existing works that assume perfect CSI for BF-PS design, we take into account the cost of channel estimation and utilize Long Short-Term Memory (LSTM) networks to design BF-PS from limited samples of estimated channel CSI. An LSTM channel extrapolator is designed first to generate high resolution estimates of the cascaded BS-RIS-user channel from sampled signals acquired at a slow time scale. Subsequently, the outputs of the channel extrapolator are fed into an LSTM-based two stage neural network for the joint design of BF-PS at a fast time scale of per coherence time. To address the critical issue that training overhead increases linearly with the number of RIS elements, we consider various pilot structures and sampling patterns in time and space to evaluate the efficiency and sum-rate performance of the proposed two time-scale design. Our results show that the proposed two time-scale design can achieve good spectral efficiency when taking into account the pilot overhead required for training. The proposed design also outperforms a direct BF-PS design that does not employ a channel extrapolator. These demonstrate the feasibility of applying RIS in time-varying channels with reasonable pilot overhead.more » « less
