Due to the interference problems of complex on-site installations attached to shield tunnel lining surface, deep learning models, developed for leakage datasets of shield tunnels, are not prepared to meet engineering requirements. Therefore, it is of utmost importance to optimize the original model based on the characteristics of leakage datasets. For this purpose, the present study adopted Mask R-CNN as the baseline and improved its performance from two aspects, including the properties of shield tunnel leakage datasets and detection errors of the original model in the testing set. With reference to the properties of leakage datasets, the model compression technique was implemented to remove the redundant parameters in the training stage and enhance the detection accuracy and speed of the original model. Besides, three error types were grouped to explore the optimization direction in practical application. Accordingly, four different optimization measurements were fulfilled step-by-step to improve the model performance. It was concluded that the compressed model with 62% sparsity reached average precision (AP) of 0.399 at the detection speed of 7 frames per second (FPS), which was 0.144 higher and 2 FPS faster than that of the original model, respectively. The improvements in the model concerned were also quite different with the three error types moderated, indicating that the optimization direction of the original model needed to be focused on the error types with higher AP promotion. In addition, the AP value of the original model significantly augmented from 25.5% to 51.1% upon minimizing the detection errors on shield tunnel leakage datasets.

The quest to answer the longstanding question of how safe is safe enough has led to an evolution of design philosophies. The traditional philosophy has been based on ensuring that system capacity exceeds demand. This philosophy has been manifested in such design methods as allowable stress design (ASD) and load and resistance factor design (LRFD). The level of safety in these methods is commonly calibrated to the acceptable practice using structural reliability methods. Such an approach to design aims to balance life safety and construction costs and works relatively well for service loads under which strength is nearly always a sufficient representative of system capacity. Under extreme loads, however, not only do factors such as ductility or energy-dissipation capacity come into play, but also the consequences of failure are vastly larger and may go beyond the failure site. For this reason, the more modern design philosophy of performance-based design emerged. According to this philosophy, a system is designed to meet a target performance, defined by limits on its performance metrics. These metrics were initially defined on structural responses, such as deformations and accelerations, and later evolved into risk measures based on monetary loss, downtime, and casualties. This philosophy is gradually making its way into modern design codes. Typical performance-based criteria focus primarily on what transpires during extreme loads. The need to incorporate the recovery of the system after such events has prompted the design philosophy to make another leap into resilience-based design (RBD). Resilience is defined herein as the ability to recover, within a predetermined period of time, in the aftermath of extreme events. Resilience transcends risk, i.e., a resilience analysis incorporates not only a risk analysis but also a recovery analysis. This paper provides a review of RBD in continuation of PBD and further introduces various applications of RBD into the design and assessment of civil infrastructure. The review covers the state of the art in resilience quantification approaches and associated computational platforms.

In this paper, a two-stage time-domain output-only damage detection method is proposed with a new energy-based damage index. In the first stage, the random decrement technique (RDT) is employed to calculate the random decrement signatures (RDSs) from the acceleration responses of a simply supported beam subjected to a moving load. The RDSs are then filtered using the Savitzky–Golay filter (SGF) in the second stage. Next, the filtered RDSs are processed by the proposed energy-based damage index to locate and quantify the intensity of the possible damage. Finally, by fitting a Gaussian curve to the damage index resulted from the non-damage conditions, the whole process is systematically implemented as a baseline-free method. The proposed method is numerically verified using a simply supported beam under moving sprung mass with different velocities and damage scenarios. The results show that the proposed method can accurately estimate the damage location/quantification from the acceleration data without any prior knowledge of either input load or damage characteristics. Additionally, the proposed method is neither sensitive to noise nor velocity variation, which makes it ideal when obtaining a constant velocity is difficult.

Predicting the way diseases spread in different societies has been thus far documented as one of the most important tools for control strategies and policy-making during a pandemic. This study is to propose a network autoregressive (NAR) model to forecast the number of total currently infected cases with coronavirus disease 2019 (COVID-19) in Iran until the end of December 2021 in view of the disease interactions within the neighboring countries in the region. For this purpose, the COVID-19 data were initially collected for seven regional nations, including Iran, Turkey, Iraq, Azerbaijan, Armenia, Afghanistan, and Pakistan. Thenceforth, a network was established over these countries, and the correlation of the disease data was calculated. Upon introducing the main structure of the NAR model, a mathematical platform was subsequently provided to further incorporate the correlation matrix into the prediction process. In addition, the maximum likelihood estimation (MLE) was utilized to determine the model parameters and optimize the forecasting accuracy. Thereafter, the number of infected cases up to December 2021 in Iran was predicted by importing the correlation matrix into the NAR model formed to observe the impact of the disease interactions in the neighboring countries. In addition, the autoregressive integrated moving average (ARIMA) was used as a benchmark to compare and validate the NAR model outcomes. The results reveal that COVID-19 data in Iran have passed the fifth peak and continue on a downward trend to bring the number of total currently infected cases below 480,000 by the end of 2021. Additionally, 20%, 50%, 80% and 95% quantiles are provided along with the point estimation to model the uncertainty in the forecast.

Structural load types, on the one hand, and structural capacity to withstand these loads, on the other hand, are of a probabilistic nature as they cannot be calculated and presented in a fully deterministic way. As such, the past few decades have witnessed the development of numerous probabilistic approaches towards the analysis and design of structures. Among the conventional methods used to assess structural reliability, the Monte Carlo sampling method has proved to be very convenient and efficient. However, it does suffer from certain disadvantages, the biggest one being the requirement of a very large number of samples to handle small probabilities, leading to a high computational cost. In this paper, a simple algorithm was proposed to estimate low failure probabilities using a small number of samples in conjunction with the Monte Carlo method. This revised approach was then presented in a step-by-step flowchart, for the purpose of easy programming and implementation.