The statistical modeling of extraordinary loads on buildings has been stagnant for decades due to the laborious and error-prone nature of existing survey methods, such as questionnaires and verbal inquiries. This study proposes a new vision-based survey method for collecting extraordinary load data by automatically analyzing surveillance videos. For this purpose, a crowd head tracking framework is developed that integrates crowd head detection and reidentification models based on convolutional neural networks to obtain head trajectories of the crowd in the survey area. The crowd head trajectories are then analyzed to extract crowd quantity and velocities, which are the essential factors for extraordinary loads. For survey areas with frequent crowd movements during temporary events, the equivalent dynamic load factor can be further estimated using crowd velocity to consider dynamic effects. A crowd quantity investigation experiment and a crowd walking experiment are conducted to validate the proposed survey method. The experimental results prove that the proposed survey method is effective and accurate in collecting load data and reasonable in considering dynamic effects during extraordinary events. The proposed survey method is easy to deploy and has the potential to collect substantial and reliable extraordinary load data for determining design load on buildings.
The objective of this study is to develop a novel and efficient model for forecasting the nonlinear behavior of structures in response to time-varying random excitation. The key idea is to design a deep learning architecture to leverage the relationships, between external excitations and structure’s vibration signals, and between historical values and future values, within multiple time-series data. The proposed method consists of two main steps: the first step applies a global attention mechanism to combine multiple-measured time series and time-varying excitation into a weighted time series before feeding it to a temporal architecture; the second step utilizes a self-attention mechanism followed by a fully connected layer to predict multi-step future values. The viability of the proposed method is demonstrated via two case studies involving synthetic data from a three-dimensional (3D) reinforced concrete structure and experimental data from an 18-story steel frame. Furthermore, comparison and robustness studies are carried out, showing that the proposed method outperforms conventional methods and maintains high performance in the presence of noise with an amplitude of less than 10%.
An increased number of hurricanes and tornadoes have been recorded worldwide in the last decade, while research efforts to reduce wind-related damage to structures become essential. Freeform architecture, which focuses on generating complex curved shapes including streamlined shapes, has recently gained interest. This study focuses on investigating the potential of kerf panels, which have unique flexibility depending on the cut patterns and densities, to generate complex shapes for façades and their performance under wind loads. To investigate the kerf panel’s potential capacity against wind loads, static and dynamic analyses were conducted for two kerf panel types with different cut densities and pre-deformed shapes. It was observed that although solid panels result in smaller displacement amplitudes, stresses, and strains in some cases, the kerf panels allow for global and local cell deformations resulting in stress reduction in various locations with the potential to reduce damage due to overstress in structures. For the pre-deformed kerf panels, it was observed that both the overall stress and strain responses in kerf cut arrangements were lower than those of the flat-shaped panels. This study shows the promise of the use of kerf panels in achieving both design flexibility and performance demands when exposed to service loadings. Considering that this newly proposed architectural configuration (design paradigm) for facades could revolutionize structural engineering by pushing complex freeform shapes to a standard practice that intertwines aesthetic arguments, building performance requirements, and material design considerations has the potential for significant practical applications.
The accurate prediction of bearing capacity is crucial in ensuring the structural integrity and safety of pile foundations. This research compares the Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), and Bidirectional LSTM (BiLSTM) algorithms utilizing a data set of 257 dynamic pile load tests for the first time. Also, this research illustrates the multicollinearity effect on DNN, CNN, RNN, LSTM, and BiLSTM models’ performance and accuracy for the first time. A comprehensive comparative analysis is conducted, employing various statistical performance parameters, rank analysis, and error matrix to evaluate the performance of these models. The performance is further validated using external validation, and visual interpretation is provided using the regression error characteristics (REC) curve and Taylor diagram. Results from the comparative analysis reveal that the DNN (Coefficient of determination (R2)training (TR) = 0.97, root mean squared error (RMSE)TR = 0.0413; R2testing (TS) = 0.9, RMSETS = 0.08) followed by BiLSTM (R2TR = 0.91, RMSETR = 0.782; R2TS = 0.89, RMSETS = 0.0862) model demonstrates the highest performance accuracy. It is noted that the BiLSTM model is better than LSTM because the BiLSTM model, which increases the amount of information for the network, is a sequence processing model made up of two LSTMs, one of which takes the input in a forward manner, and the other in a backward direction. The prediction of pile-bearing capacity is strongly influenced by ram weight (having a considerable multicollinearity level), and the effect of the considerable multicollinearity level has been determined for the model based on the recurrent neural network approach. In this study, the recurrent neural network model has the least performance and accuracy in predicting the pile-bearing capacity.
Shield tunnel lining is prone to water leakage, which may further bring about corrosion and structural damage to the walls, potentially leading to dangerous accidents. To avoid tedious and inefficient manual inspection, many projects use artificial intelligence (AI) to detect cracks and water leakage. A novel method for water leakage inspection in shield tunnel lining that utilizes deep learning is introduced in this paper. Our proposal includes a ConvNeXt-S backbone, deconvolutional-feature pyramid network (D-FPN), spatial attention module (SPAM). and a detection head. It can extract representative features of leaking areas to aid inspection processes. To further improve the model’s robustness, we innovatively use an inversed low-light enhancement method to convert normally illuminated images to low light ones and introduce them into the training samples. Validation experiments are performed, achieving the average precision (AP) score of 56.8%, which outperforms previous work by a margin of 5.7%. Visualization illustrations also support our method’s practical effectiveness.
Complex modulus (G*) is one of the important criteria for asphalt classification according to AASHTO M320-10, and is often used to predict the linear viscoelastic behavior of asphalt binders. In addition, phase angle (φ) characterizes the deformation resilience of asphalt and is used to assess the ratio between the viscous and elastic components. It is thus important to quickly and accurately estimate these two indicators. The purpose of this investigation is to construct an extreme gradient boosting (XGB) model to predict G* and φ of graphene oxide (GO) modified asphalt at medium and high temperatures. Two data sets are gathered from previously published experiments, consisting of 357 samples for G* and 339 samples for φ, and these are used to develop the XGB model using nine inputs representing the asphalt binder components. The findings show that XGB is an excellent predictor of G* and φ of GO-modified asphalt, evaluated by the coefficient of determination R2 (R2 = 0.990 and 0.9903 for G* and φ, respectively) and root mean square error (RMSE = 31.499 and 1.08 for G* and φ, respectively). In addition, the model’s performance is compared with experimental results and five other machine learning (ML) models to highlight its accuracy. In the final step, the Shapley additive explanations (SHAP) value analysis is conducted to assess the impact of each input and the correlation between pairs of important features on asphalt’s two physical properties.
Flexural performance of joints is critical for prefabricated structures. This study presents a novel channel steel-bolt (CB) joint for prefabricated subway stations. Full-scale tests are carried out to investigate the flexural behavior of the CB joint under the design loads of the test-case station. In addition, a three dimensional (3D) finite element (FE) model of the CB joint is established, incorporating viscous contact to simulate the bonding and detachment behaviors of the interface between channel steel and concrete. Based on the 3D FE model, the study examines the flexural bearing mechanism and influencing factors for the flexural performance of the CB joint. The results indicate that the flexural behavior of the CB joint exhibits significant nonlinear characteristics, which can be divided into four stages. To illustrate the piecewise linearity of the bending moment-rotational angle curve, a four-stage simplified model is proposed, which is easily applicable in engineering practice. The study reveals that axial force can enhance the flexural capacity of the CB joint, while the preload of the bolt has a negligible effect. The flexural capacity of the CB joint is approximate twice the value of the designed bending moment, demonstrating that the joint is suitable for the test-case station.
Traditional asphalt concrete (AC) and stone matrix asphalt (SMA), which are used as thin asphalt overlays, are common maintenance strategies to enhancing ride quality, skid resistance, and durability. Recently, several studies have used a novel asphalt mixture known as a high-friction thin overlay (HFTO) to improve surface characteristics. However, it remains uncertain whether the laboratory properties of HFTO differ significantly from those of conventional mixtures. This study aims to evaluate the laboratory properties of HFTO mixtures and compare them with those of AC and SMA. Those mixtures with nominal maximum size of 9.5 mm were produced in the laboratory, and performance tests were conducted, including wheel tracking test, low temperature flexural creep test, moisture susceptibility test, Cantabro Abrasion Test, Marshall Test, sand patch test, British pendulum test, and indoor tire-rolling-down test. The results showed that the HFTO exhibited a lower tire/pavement noise than the AC and SMA. Additionally, HFTO had superior high-temperature stability, larger macro texture, and higher skid resistance in comparison to those of AC, but lower than those of SMA. Consequently, HFTO mixtures may be considered a suitable replacement for traditional AC mixtures in regions where skid resistance and noise reduction are concerns.
Raveling is a common distress of asphalt pavements, defined as the removal of stones from the pavement surface. To predict and assess raveling quantitatively, a cumulative damage model based on an energy dissipation approach has been developed at the meso level. To construct the model, a new test method, the pendulum impact test, was employed to determine the fracture energy of the stone-mastic-stone meso-unit, while digital image analysis and dynamic shear rheometer test were used to acquire the strain rate of specimens and the rheology property of mastic, respectively. Analysis of the model reveals that when the material properties remain constant, the cumulative damage is directly correlated with loading time, loading amplitude, and loading frequency. Specifically, damage increases with superimposed linear and cosine variations over time. A higher stress amplitude results in a more rapidly increasing rate of damage, while a lower load frequency leads to more severe damage within the same loading time. Moreover, an example of the application of the model has been presented, showing that the model can be utilized to estimate failure life due to raveling. The model is able to offer a theoretical foundation for the design and maintenance of anti-raveling asphalt pavements.
