Search

• A response spectrum to predict structural responses under crowd jumping loads is derived by stochastic vibration theory. • The closed-form expression of the response spectrum is established. • The response spectrum is verified by measurements of a real floor. A major challenge in the vibration performance assessment of structures under crowd jumping loads is the lack of efficient prediction models. With the challenge in mind, this study develops a multi-mode response spectrum for multi-harmonic crowd jumping loads to predict the root mean square value of structural accelerations according to the stochastic vibration theory. The response spectrum is applicable to structures with arbitrary vibration modes, because crowd jumping loads are regarded as multi-point excitations. A random field model proposed by the author is used to derive the closed-form expression of the single-mode response induced by individual jumping loads of a single-harmonic. Furthermore, the closed-form expressions of the correlation coefficients for different load harmonics, different vibration modes, and different jumping persons are established based on the random field model. The performance of the response spectrum is evaluated for two floors, and the results indicate the feasibility of the proposed approach to predicting structural accelerations subjected to crowd jumping loads. [ABSTRACT FROM AUTHOR]

Features of jumping loads are essentially high-dimensional random variables but have been simplistically modeled owing to the lack of proper mathematical tools. Generative adversarial networks (GANs) in conjunction with deep learning technology are adopted herein for modeling the jumping loads. Conditional GANs (CGANs) combined with Wasserstein GANs (WGANs) with gradient penalty (WGANs-GP) are adopted in the impulse modeling, where a multi-layer perceptron and a convolutional neural network are employed for the discriminator and generator, respectively. As for the impulse amplitude sequence and interval sequence modeling, similar CGANs combined with WGANs-GP are adopted, where recurrent neural networks are employed for both the generator and discriminator. A large amount of measured individual jumping loads are utilized in training GANs to ensure the generated samples can simulate the real ones well. After training, the individual jumping loads are simulated by connecting the generated impulse samples, based on the generated impulse amplitude sequence samples and interval sequence samples. The simulated jumping loads can be used to assess the vibration performance of assembly structures, such as grandstands, concert halls, and gym floors. Moreover, the established GANs can be extended to the modeling of other stochastic dynamic excitations. [ABSTRACT FROM AUTHOR]

This paper reports a dataset that provides force time histories of various types of human-induced loads, including walking, jumping, and bouncing activities guided by metronome or music. The dataset contains 3,518 records collected from six experimental campaigns. Ground reaction forces of test subjects were measured using high-precision force plates for individuals and small groups, and the displacements of certain body parts of the test subjects were measured using three-dimensional motion capture technology for large crowds. All data records were converted to a unified format and uploaded to a published dataset. An accompanying website of the human-induced load dataset has also been designed and put online. The website provides a publicly available and type-classified data source of human-induced loads. The dataset will be updated by adding new measurements from both the authors and other researchers. To the author's knowledge, this dataset is the first open access dataset for research on human-induced loads, which is of great significance to future studies on new and advanced load models to understand the crowd load mechanisms under different circumstances and to spark broader data sharing in this area. Moreover, this dataset helps design engineers select real human-induced loads for a time history analysis of structures and fosters future collaboration among researchers and engineers. [ABSTRACT FROM AUTHOR]

• The equation of motion of a crowd-structure coupled system is established. • Structural acceleration responses are derived by stochastic vibration theory. • Two tests on a bridge model and a real floor are conducted to verify the methods. Modern slender structures such as cantilever stands and long-span floors are quite sensitive to vibrations induced by crowd jumping. Existing human-structure interaction models cannot consider human body parameter variabilities and apply to jumping-induced vibration analysis of structures with closely-spaced and complex vibration modes and passive crowd. This paper herein aims to derive the equation of motion and its spectral analysis solution of an advanced model for the crowd-structure coupled system. Each person on a structure is treated as a separate spring-mass-damper, and the structure is treated as a multi-degree-of-freedom system. Each person is classified as active crowd or passive crowd according to whether he/she exerts dynamic forces on the structure. Then, the equation of motion of the crowd-structure coupled system is derived, and its spectral analysis solution is established by the stochastic vibration theory and the state space method. The frequency response function is adopted to analyze dynamic properties of the coupled system. With a previously-proposed random field model for crowd jumping loads, vibration responses of the coupled system are analyzed. Comparisons between the simulations and the results from two experiments demonstrate the feasibility of the proposed spectral analysis method. [ABSTRACT FROM AUTHOR]

Severe vibrations may occur on slender structures like footbridges and cantilever stands due to human-induced loads such as walking, jumping or bouncing. Currently, to develop a load model for structural design, the main features, such as periodicity and stationarity of experimental load records, are artificially extracted and then mathematically modeled. Different physical features have been included in different load models, i.e. no unified load model exists for different individual activities. The recently emerged generative adversarial networks can be used to model high-dimensional random variables. The probability distribution of these variables learned from real samples can be used to generate new samples, avoiding extracting features artificially. In this paper, a new model is proposed which combines the conditional generative adversarial networks and Wasserstein generative adversarial networks with gradient penalty to generate individual walking, jumping and bouncing loads. The generator of the model has five fully connected layers and a one-dimensional convolutional layer, and the discriminator has five fully connected layers. After one million training steps using the experimental records, the generator can generate high-quality samples similar to real samples in waveform. Finally, by comparing the power spectral densities and single degree of freedom system's responses of the generated samples with real samples, it is further proved that the proposed generative adversarial network model can be used to simulate various human-induced loads. Source code of the model along with its trained weights is provided to the readers to further analysis and application. [ABSTRACT FROM AUTHOR]

Highlights • A random field model for crowd jumping load is proposed. • Experiments are conducted on a jumping group of 48 persons using 3D MCT. • Structural responses are predicted by the stochastic vibration theory. Abstract A reliable model for crowd jumping loads is a prerequisite to an accurate prediction of a long-span structure's response when subjected to crowd motion triggered by music and other mass appeal events. Because of the lack of crowd jumping experimental data, crowd jumping load models are scarce with few data available on this type of synchronisation among people. Inspired by the random field model of earthquake excitations, this study attempts to develop a random field model for crowd jumping loads based on extensive records from unique crowd jumping load experiments. A framework for the random field model of crowd jumping is first proposed, in which auto-power spectral density (PSD) and cross-PSD functions of individual and group jumping loads are defined. The auto-PSD was obtained by updating a model in a previous study. Experiments were conducted on a jumping group of 48 persons using a 3D motion capture technology. A cross-PSD model for any two persons in a jumping group was then developed based on the experimental data. Having obtained a PSD matrix for crowd jumping, the structural response could then be predicted by the stochastic vibration theory. Finally, the feasibility of the model was verified by comparing the measured responses of an existing floor with the prediction responses using the proposed model. [ABSTRACT FROM AUTHOR]

Modern slender structures such as long-span floors and cantilever stands are sensitive to jumping-induced vibrations. A conventional deterministic Fourier series model for the human jumping load may overestimate a structure's responses in resonance condition. This paper suggests a power spectral density (PSD) function for the individual jumping load, which was treated as a narrowband stationary stochastic process. Experiments were conducted on individual jumping loads resulting in 334 records from 73 subjects. Statistical analysis of the records led to experimental PSD curves on which a symmetrical bilinear function was suggested. The proposed PSD function is centered on the given jumping frequency and its integer multiples. The function's parameters were determined by equating the total energy of the proposed PSD with that of the experimental PSD. Application of the proposed PSD for predicting a floor's peak response via the stochastic vibration theory was then presented. The predictions for an experimental floor model subjected to individual jumping were compared with the measured peak responses. The comparisons demonstrated that the proposed PSD function was applicable for predicting the floor's response to individual jumping. Finally, the framework for calculating crowd-induced structure vibrations using the suggested PSD was also discussed. [ABSTRACT FROM AUTHOR]

Offshore structures, such as oil and gas platforms and offshore wind turbines, are subjected to wind and wave loads simultaneously during their service lifetime. Since the wind and wave states are of significant randomness and dependence, the probabilistic modeling of joint wind and wave conditions plays an essential role in the safety design of offshore structures. Currently, three different methods can be adopted to establish the joint probabilistic model, which, however, are somewhat inconvenient in applications. The recently emerged generative adversarial networks has been demonstrated to be effective in dealing with high-dimensional random variables in several fields. In this study, the implicit joint probabilistic model of joint wind and wave load conditions is developed based on the Wasserstein generative adversarial network with gradient penalty. Long-term metocean reanalysis data of the site in the South China Sea is used to train and validate the model. After one million training steps, high-quality samples that are quite similar to the original data can be generated by the developed model. In addition, statistical comparisons of the generated samples obtained by the C-vine copula approach and the developed generative adversarial network model are performed as well, which demonstrates the effectiveness and superiority of the developed model. [ABSTRACT FROM AUTHOR]

• Macroscopic uncoordinated deformation processes were studied combining DIC. • Microscopic failure mechanisms were studied through PFC3D numerical simulations. • Damage constitutive model considering WIZ thickness and dip angle was established. The weak interlayer zone (WIZ) is a high-risk geotechnical system with poor mechanical properties, strong ductility, loose structure, and wide distribution. To comprehensively investigate the uncoordinated deformation characteristics and failure mechanism of rock mass with WIZ, a series of uniaxial loading tests using digital image correlation (DIC) technique, and numerical simulation tests were carried out. Besides, the damage constitutive model of rock mass with WIZ was established, then the damage evolution law was also analyzed. Results show that a high strain localized zone appears in the WIZ from the beginning of loading, which then develops towards the contact surface between the WIZ and the rock mass. As loading proceeds, the WIZ exhibits layered spalling and ultimately forms through-cracks after the initiation of micro-cracks, and vertical tensile cracks to appear in the rock. causing vertical tensile cracks to appear in the rock. With the increase of the WIZ thickness and dip angle, the uniaxial compressive strength and deformation modulus of specimen show a negative exponential decay. When rock mass with WIZ collapse, the bonding of WIZ particles fails and the number of tensile-shear cracks gradually increases, eventually resulting in the plastic squeezing-out of WIZ. Moreover, the particles of the rock masses with smaller WIZ dip angles are mainly displaced vertically, and the failure is mainly caused by tensile splitting. The significant lateral displacement of particles in the rock mass with larger WIZ dip angles leads to the destruction of the rock mass mainly through shear slip, in particular a pronounced sliding down along the WIZ. Furthermore, the damage constitutive model of the rock mass with WIZ was established by distinguishing the total damage of the rock mass with WIZ into the damage of the WIZ and the damage of the rock mass. The model prediction results show that the total damage D increases with the increase of WIZ thickness and dip angle, and develops faster in the early stage and slower in the later stage. The research can provide effective basis and feasible ideas for the uncoordinated deformation and failure evolution, macro–micro failure mechanism, as well as the constitutive model of rock mass with WIZ in underground cavern under high geostress. [ABSTRACT FROM AUTHOR]

Strong mainshocks are typically accompanied by numerous aftershocks, and the investigation of the structural failure mechanisms under the mainshock-aftershock sequence becomes particularly crucial. However, the number of recorded mainshock-aftershock sequences is limited. Therefore, the purpose of this article is to provide a reasonable method for directly generating the aftershock time histories from mainshock time histories. Using convolutional network as the basic network layer and conditional generative adversarial network as the structure, two models, one-dimensional convolution (1D-C-DCGAN) and two-dimensional convolution (2D-C-DCGAN) are established respectively by utilizing the deep convolutional generative adversarial network to learn the relationship between the mainshock-aftershock time histories. Then, they are trained with 972 pairs of the selected mainshock-aftershock time histories, and prediction results are discussed in comparison. The results show that the two models are proficient in generating AS acceleration time histories that are closely related to the sample trend, in which the 2D-C-DCGAN model performing better in overall waveform prediction, but with local spikes. In the comparison of intensity measures and response spectra, by examining coefficients such as R2, RMSE, MAPE, the two models outperformed the mainstream model (ASK14) on the dataset, and the 2D-C-DCGAN model is more accurate than the 1D-C-DCGAN model. The distributions of intensity measure predicted by 2D-C-DCGAN model are closer to the measured intensity measures, and its predicted response spectra are smoother and better matched to the measured response spectra. This advantage can be attributed to the effectiveness of convolution operations on two-dimensional data, allowing the convolutional capabilities to be fully utilized. [ABSTRACT FROM AUTHOR]

Strong mainshocks are typically accompanied by numerous aftershocks, and the investigation of the structural failure mechanisms under the mainshock-aftershock sequence becomes particularly crucial. However, the number of recorded mainshock-aftershock sequences is limited. Therefore, the purpose of this article is to provide a reasonable method for directly generating the aftershock time histories from mainshock time histories. Using convolutional network as the basic network layer and conditional generative adversarial network as the structure, two models, one-dimensional convolution (1D-C-DCGAN) and two-dimensional convolution (2D-C-DCGAN) are established respectively by utilizing the deep convolutional generative adversarial network to learn the relationship between the mainshock-aftershock time histories. Then, they are trained with 972 pairs of the selected mainshock-aftershock time histories, and prediction results are discussed in comparison. The results show that the two models are proficient in generating AS acceleration time histories that are closely related to the sample trend, in which the 2D-C-DCGAN model performing better in overall waveform prediction, but with local spikes. In the comparison of intensity measures and response spectra, by examining coefficients such as R2, RMSE, MAPE, the two models outperformed the mainstream model (ASK14) on the dataset, and the 2D-C-DCGAN model is more accurate than the 1D-C-DCGAN model. The distributions of intensity measure predicted by 2D-C-DCGAN model are closer to the measured intensity measures, and its predicted response spectra are smoother and better matched to the measured response spectra. This advantage can be attributed to the effectiveness of convolution operations on two-dimensional data, allowing the convolutional capabilities to be fully utilized. [ABSTRACT FROM AUTHOR]

• A database has been established by collecting 204 sets of engineering cases. • A multi-factor-coupling geostress model was established using AutoGluon. • The model exhibits higher accuracy compared to traditional criteria. Accurate evaluation of geostress levels is crucial for the investigation, design, and construction of underground engineering. However, existing methods for geostress classification are limited by their reliance on singular indexes, leading to discrepancies between the evaluation results and the actual observations in engineering. The discrepancies have resulted in casualties and economic losses. To address the above issues, 204 sets of geostress levels data were collected from numerous documents in the study. Six indexes related to geostress levels have been identified by analyzing the collected data and traditional criteria. Additionally, an orthogonal experimental design was carried out on the range-based indexes to obtain a multi-index fusion database. Furthermore, an intelligent evaluation model for the geostress levels was established in combination with the AutoGluon automatic machine learning framework. Subsequently, software has been developed from this model for ease of use in practical engineering. After on-site data verification, the proposed model demonstrated an accuracy of 95%, outperforming five traditional criteria, with the highest accuracy among them being 70%. The research could provide an effective basis for geostress classification, thereby enhancing the safety of on-site construction personnel. [ABSTRACT FROM AUTHOR]

• WIZ constitutive model based on modified plastic work with stress path independency. • Deterioration characteristics of mechanical parameters of WIZ with modified plastic work. • Accurate representation of non-linear behavior and dilation in WIZ by the constitutive model. A simple yet innovative constitutive model of the weak interlayer zone (WIZ) is proposed to represent the unloading mechanical response and the failure mechanism of WIZ under complex unloading stress paths in underground excavation based on credible experimental results. The established model employs a modified equivalent plastic work with obvious stress path independency as the hardening parameter to appropriately interpret the effect of stress paths on the unloading plastic strain hardening characteristics of WIZ. Moreover, the deterioration of mechanical parameters with the hardening parameter is simultaneously explored. Additionally, the modified equivalent plastic work hardening constitutive model has been created with the aid of the updated Mogi–Coulomb yield criterion, associated flow rule for tensile failure, and non-associated flow rule for shear failure, in order to accurately characterize the non-linear properties and significant dilation throughout the entire unloading conditions. The capability and rationality of the established model were then investigated and verified through the finite difference numerical algorithm by comparing several numeric examples with the laboratory unloading triaxial test results and the field deformation monitoring data. The model was then used to simulate and analyze the stability of rock masses with WIZ during the large underground cavern excavations. Results show that the model can effectively characterize the stress redistribution, the large discontinuous deformation of upper and lower rock masses of WIZ, as well as the structural stress-induced failures. The study can offer a crucial scientific foundation for appropriate construction and support optimization of underground engineering affected by WIZ. [ABSTRACT FROM AUTHOR]

• An improved YOLOv5 model for tunnel crack detection is proposed. • Model overcomes low light, low contrast and high noise challenges in tunnels. • Improved model achieves 90% detection accuracy while realizing real-time detection. • Improved model can be applied to practical engineering through field validation. An improved real-time tunnel lining crack detection model based on YOLOv5 is proposed. This model maintains high precision and accuracy crack detection in low-light, low-contrast and high-noise environments by introducing several effective data augmentation techniques as well as semantic context encoding (SCE) and detail preserving encoding (DPE) at the head of the network structure. It achieves 90 % precision, 91 % recall, and 92 % mAP@50. The model demonstrates better detection performance than YOLOv4-tiny, YOLOv5s, YOLOv8s, and traditional threshold segmentation method, especially in complex environments to reduce misdetection and omission. The average detection time is only 12 ms per image, demonstrating the feasibility of its real-time application. The robust and generalization performance of the model is validated in specific engineering applications, showing great potential for improving detection efficiency, cost-effectiveness, and reliability in tunnel safety assessment and disasters management. [ABSTRACT FROM AUTHOR]

With the escalating development and utilization of subterranean spaces, the seismic hazards faced by underground structures are progressively increasing. However, owing to the challenges associated with acquiring underground seismic data and historical seismic design norms, pertinent regulations and research in this domain are scarce. This study focused on three crucial intensity measures in the seismic design process of underground structures peak ground acceleration (PGA), peak ground velocity (PGV), and peak ground displacement (PGD). The research leverages seismic data obtained from the California Strong Motion Instrumentation Program (CSMIP) to train and evaluate a conditional generative adversarial network (CGAN) model. This model was employed to establish a multivariate joint conditional probability distribution among the intensity measures at varying depths, facilitating the stochastic prediction of shallow intensity measures. In contrast to empirical formulas, the CGAN model eliminates the need for a predefined equation structure and enables the simultaneous prediction of multiple intensity measures. The performance of the model was evaluated by comparing the predictive accuracy of the CGAN model and empirical fitting formulas across diverse site conditions and depth intervals using metrics such as relative error coefficients. It can be concluded that the proposed CGAN model can accurately predict shallow seismic intensity measures, and the predictions conform to a specific conditional distribution while retaining the stochastic nature of seismic motion. Compared with empirical formula models, the CGAN model exhibited an enhanced predictive capability. • CGAN model extracts inter-depth seismic intensity measures distribution. • Predictions of the CGAN model are more accurate than empirical formula model. • CGAN model outperforms for larger seismic intensity measures. • CGAN model exhibits varied performance with site and depth differences. [ABSTRACT FROM AUTHOR]

This study delves into the mechanical behavior of interlayer soils in conventional French railway track beds. The focus is on the influence of four key factors: volumetric coarse grains content, stress state, number of cycles, and water content. Comprehensive analysis of experimental results reveals that permanent deformation and resilient modulus are significantly affected by the interplay of these factors. Notably, the relationship between these factors and the permanent deformation of interlayer soils is sophisticated and coupled, and more advanced models may be required to sufficiently reflect their behavior. To better understand the complex relationships among these factors and accurately predict the behavior of interlayer soil, a fatigue model based on Artificial Neural Networks (ANN, i.e. a classical data-driven approach) was developed. The model demonstrates high prediction reliability and accuracy, with a determination coefficient (R2) of 0.9996, a mean absolute error (MAE) of 0.0044, and a root mean square error (RMSE) of 0.006629. Thereafter, the proposed model was compared with the laboratory cyclic test results as well as with the commonly-used empirical fatigue models, and a permanent strain curve of the interlayer soil were also successfully predicted by using test set. Results show that the proposed model could effectively capture the influence of multiple coupled factors on permanent plastic strain, adapting to a wide range of interlayer soil conditions. In conclusion, the data-driven fatigue model provides valuable insights into the combined effects of various factors on interlayer soil behavior, offering an effective tool for evaluating the performance of French traditional track beds under different conditions. [ABSTRACT FROM AUTHOR]

• A simplified analytical model based on equivalent hydraulic radius was proposed. • The distribution law of seepage pressure of non-circular tunnel could be derived. • Accuracy and superiority of the equivalent hydraulic radius method were verified. • Parameters of each part of the composite lining were optimized. • The failure of the lining before and after parameter optimization was analyzed. The water pressure distribution acting on non-circular hydraulic tunnels have always been an important and crucial issue in deep-buried tunnels and caverns, due to its complexity, variability and potential threats to the stability of surrounding rock and the safety of lining structure. This research proposes an accurate axisymmetric analytical solution for water pressure and seepage discharge on the surface of surrounding rock, the lining and grouting circle under high external and internal water pressures, by introducing an equivalent hydraulic radius method in to seepage theory in fractured media. Then, the distribution law of seepage pressure within different distances from the tunnel wall is derived by theoretical analysis. The accuracy, superiority and applicability of the theoretical solution based on the equivalent hydraulic radius are verified sufficiently. The maximum errors of the theoretical solution are less than 15%, 18% and 5.98% respectively, when compared to the numerical solution, the in-situ water pressure monitoring data and the theoretical solution based on the conformal mapping. And the proposed theoretical solution is much closer to the numerical solution than the existing equivalent methods. Moreover, parameters of each part of the composite lining are optimized. Increasing 10% of the original of the permeability coefficient of the composite lining can maximally reduce the external water pressure, and the maximum principal stress and displacement of the composite lining can be effectively reduced by comparing the contour of the maximum principal stress and displacement before and after parameter optimization. Furthermore, by comparing the excavation damage zone distribution before and after parameter optimization, the shear failure occurring in the composite lining is significantly reduced, and the stability of the composite lining structure can be improved. The conclusions could provide important guiding theoretical value for the further study of water pressure distribution acting on non-circular hydraulic tunnels as well as for the correct grouting and drainage designs. [ABSTRACT FROM AUTHOR]