Advances in Engineering Software最新文献

Recent refactoring of the GPUSPH codebase have uncovered some of the limitations of the official CUDA compiler (nvcc) offered by NVIDIA when dealing with some C++ constructs, which has shed some new light on the relative importance of the neighbors list construction and traversal in SPH codes, presenting new possibility of optimization with surprising performance gains. We present our solution for high-performance neighbors list construction and traversal, and show that a $4\times$ speedup can be achieved in industrial applications.

The use of fracture energy regularization techniques can effectively mitigate the mesh dependency of numerical solutions caused by the strain softening behavior of quasi-brittle materials. However, the successful regularization depends on the correct estimation of the crack bandwidth in Finite Element solutions. This paper aims to present an enhanced crack band formulation to overcome the strain localization instability especially for the higher-order elements developed in the framework of Carrera Unified Formulation (CUF). Besides, a modified Mazars damage method incorporating fracture energy regularization is employed to describe the nonlinear damage behavior of the concrete. To evaluate the efficiency of the proposed crack band formulation, three experimental concrete benchmarks are selected for the numerical damage analysis. By comparing numerical and experimental results, the proposed method can guarantee mesh objectivity despite varying finite element numbers and orders, indicating perseved fracture energy consumption within proposed higher-order beam models.

With recent trends reducing the carbon footprint of concrete, more novel materials are designed. It's mostly done by replacing cement with admixtures that are wastes in industrial processes. There is a need to provide reliable and accurate models to estimate the properties of the material. In this case the selected ML algorithms such as ANN, RF and DT were used for estimating the pull-off strength of the surface layer of cement mortar containing granite powder, fly ash and ground granulated blast furnace slag. The focus was on the cement-sand ratio of 1:3, replacing up to 30 % of the binder. Ultrasonic pulse velocity and pull-off strength of the surface layer. The analyses were performed in comparative manner and proved the accuracy of the designed models. The error values (MAPE, NRMSE and MAE) of the most effective model is below 3,5 %, indicating an extremely high success rate in prediction. An R² ratio of 0.9436 confirms the very good fit of the model. Parametric tests were performed and SHAP analysis gave a better understanding of the models. The main conclusion of the study is to identify the possibility of replacing destructive testing with non-destructive testing supported by machine learning and material information to determine the pull-off strength of the subsurface layer at a selected depth for cement mortars containing waste materials. A particular advantage of the presented approach is the possibility of reducing the time to determine selected desired material parameters and the amount of testing required compared to the traditional approach.

Bridge photographs contain significant technical information, such as damaged structural parts and types of damage, yet interpreting these details is not always straightforward. Despite the advancements in image analysis for bridge inspection, there remains a significant gap in converting these images into comprehensible explanatory texts that can be readily used by less experienced engineers and administrative staff for effective maintenance decision-making. In this study, we developed a model that generates explanatory texts from bridge images based on a deep learning model, and we also developed a web system that can be utilized during bridge inspections. The proposed method enables the provision of user-friendly, text-based explanations of bridge damage within images, allowing relatively inexperienced engineers and administrative staff without extensive technical expertise to understand the representation of bridge damage in text form. Additionally, we have developed a system that continually trains and improves its performance by accumulating data as users interact with it. This paper describes the image captioning technique for generating explanatory texts and the structure of the web system.

Soil-embedded vehicle barrier systems are frequently placed along high-speed highways to safely redirect errant motorists away from roadside hazards. Improved knowledge and understanding of the dynamic interactions between posts and soil are essential for advancing and optimizing these protective systems. Although the Finite Element Method (FEM) is a standard tool in the design, analysis, and evaluation of such systems, its conventional application faces challenges in accurately simulating the large soil deformations encountered by post-soil systems under impact loading. In this study, we introduce an innovative computational framework designed to simulate dynamic post-soil interactions through an adaptive coupling of the FEM and Smoothed Particle Hydrodynamics (SPH). The adaptive FEM-SPH approachʼs accuracy was validated through quantitative and qualitative analyses, benchmarked against empirical data from a unique series of physical impact tests. The results from the adaptive FEM-SPH model demonstrated remarkable agreement with observed force vs. displacement and energy vs. displacement responses, emphasizing its potential as a viable tool for assessing the performance and behavior of post-soil systems under vehicular impacts. Comparative analysis with existing simulation techniques for addressing the post-soil impact problem highlighted the adaptive FEM-SPH model's adaptability, robustness, and accuracy, thereby enriching the understanding of dynamic soil-structure interactions under impact loading. Moreover, this approach facilitated the derivation of a unique relationship between the post's center of rotation and its embedment depth, offering valuable insights for designing and optimizing barrier systems. The implications of our findings are poised to augment the design, analysis, and overall effectiveness of barrier systems, contributing to enhanced motorist safety.

This paper presents an approach for the topology optimization problem with pressure load. The approach is constructed by combining Moving Morphable Void (MMV) approach with Boundary Element Method (BEM). In this approach, the pressure boundary is explicitly described using B-spline curves and optimized simultaneously with free boundary. In the current approach, not only the moving load boundary is traced without any predefined identification scheme, but also the pressure load can be applied accurately to the structure without any needs for special load interpolation scheme. Several numerical examples in two dimensions are explored to demonstrate the effectiveness and advantages of the present approach.

In this paper we introduce GP+, an open-source library for kernel-based learning via Gaussian processes (GPs) which are powerful statistical models that are completely characterized by their parametric covariance and mean functions. GP+ is built on PyTorch and provides a user-friendly and object-oriented tool for probabilistic learning and inference. As we demonstrate with a host of examples, GP+ has a few unique advantages over other GP modeling libraries. We achieve these advantages primarily by integrating nonlinear manifold learning techniques with GPs’ covariance and mean functions. As part of introducing GP+, in this paper we also make methodological contributions that $\left(1\right)$ enable probabilistic data fusion and inverse parameter estimation, and $\left(2\right)$ equip GPs with parsimonious parametric mean functions which span mixed feature spaces that have both categorical and quantitative variables. We demonstrate the impact of these contributions in the context of Bayesian optimization, multi-fidelity modeling, sensitivity analysis, and calibration of computer models.

This study presents a hybrid model coupling particle swarm optimization (PSO) with group method of data handling (GMDH) for predicting the ultimate strength of rectangular concrete-filled steel tube (RCFST) columns. A large database of 490 data samples collected from the existing literature was used to construct the model. Compared with the optimal model among the nine existing models, the coefficient of variation (COV), mean absolute percentage error (MAPE) and root relative squared error (RRSE) values of all datasets of the PSO-GMDH model were decreased by 58.38 %, 69.22 % and 64.27 %, respectively; while the coefficient of determination (R²) and a20-index values were increased by 34.32 % and 8.65 %, respectively. The results show that the predicted results of PSO-GMDH model are in good agreement with the experimental results and can accurately predict the ultimate strength of rectangular RCFST columns. In addition, a graphical user interface (GUI) has been developed to facilitate the application of the PSO-GMDH model.

In this paper, we innovatively propose the Arctic Puffin Optimization (APO), a metaheuristic optimization algorithm inspired by the survival and predation behaviors of the Arctic puffin. The APO consists of an aerial flight (exploration) and an underwater foraging (exploitation) phase. In the exploration phase, the Levy flight and velocity factor mechanisms are introduced to enhance the algorithm's ability to jump out of local optima and improve the convergence speed. In the exploitation phase, strategies such as the synergy and adaptive change factors are used to ensure that the algorithm can effectively utilize the current best solution and guide the search direction. In addition, the dynamic transition between the exploration and development phases is realized through the behavioral conversion factor, which effectively balances global search and local development. In order to verify the advancement and applicability of the APO algorithm, it is compared with nine advanced optimization algorithms. In the three test sets of CEC2017, CEC2019, and CEC2022, the APO algorithm outperforms the other compared algorithms in 72%, 70%, and 75% of the cases, respectively. Meanwhile, the Wilcoxon signed-rank test results and Friedman rank-mean statistically prove the superiority of the APO algorithm. Furthermore, on thirteen real-world engineering problems, APO outperforms the other compared algorithms in 85% of the test cases, demonstrating its potential in solving complex real-world optimization problems. In summary, APO proves its practical value and advantages in solving various complex optimization problems by its excellent performance.

In this paper, a new meta-heuristic optimization algorithm motivated by the foraging behaviour of blood-sucking leeches in rice fields is presented, named Blood-Sucking Leech Optimizer (BSLO). BSLO is modelled by five hunting strategies, which are the exploration of directional leeches, exploitation of directional leeches, switching mechanism of directional leeches, search strategy of directionless leeches, and re-tracking strategy. BSLO and ten comparative meta-heuristic optimization algorithms are used for optimizing twenty-three classical benchmark functions, CEC 2017, and CEC 2019. The strong robustness and optimization efficiency of BSLO are confirmed via four qualitative analyses, two statistical tests and convergence curves. Furthermore, the superiority of BSLO for real-world problems under constraints is demonstrated using five classical engineering problems. Finally, a BSLO-based Artificial Neural Network (ANN) predictive model for diameter prediction of melt electrospinning writing fibre is proposed, which further verifies BSLO's applicability for real-world problems. Therefore, BSLO is a potential optimizer for optimizing various problems. Source codes of BSLO are publicly available at https://www.mathworks.com/matlabcentral/fileexchange/163106-blood-sucking-leech-optimizer.