Search for: All records

Abstract Glass fiber-reinforced polymers (GFRP) are widely applied to enhance the strength of concrete columns due to their lightweight and high-strength characteristics. This study presents the development of a metaheuristics-guided machine learning (ML) model for predicting the compressive strength (CS) of GFRP-confined concrete columns (GFRP-CC). Traditional predictive models, primarily based on Linear or nonlinear regression, are often limited by narrow data scopes and methodological constraints. To address this gap, we propose an innovative ML model, leveraging an extensive database of 319 experimental results compiled from 41 peer-reviewed articles spanning 1991–2024. Using an artificial neural network (ANN) combined with five metaheuristic algorithms, the study aims to reduce the dependency on costly and time-intensive laboratory testing. The model development considered eight key parameters: diameter of the compression member (D), height of the compression member (H), compressive strength of unconfined concrete (f′_co), GFRP reinforcement ratio (ρ_f), tensile modulus of elasticity of GFRP (E_f), ultimate tensile strength of GFRP (f_f), nominal thickness of GFRP reinforcement (t_f), and number of GFRP layers. Among the tested models, the Stochastic Paint Optimizer (SPO)-ANN model demonstrated the highest accuracy, achieving a coefficient of determination of 0.9630 with minimal error values. To ensure transparency and interpretability, SHapley Additive exPlanations (SHAP), Olden methodologies, and Partial dependence were employed to elucidate the relative importance of contributing features. Critical factors influencing the CS of GFRP-CC included the thickness of GFRP reinforcement, tensile strength, and layer count. A user-friendly graphical interface was developed to facilitate practical adoption, enabling researchers and practitioners to model CFRP-CC compressive strength efficiently. This work represents a paradigm shift in concrete research, emphasizing sophisticated, data-driven methodologies that bridge the gap between experimental data and practical applications. 

Vehicles controlled by autonomous driving software (ADS) are expected to bring many social and economic benefits, but at the current stage not being broadly used due to concerns with regard to their safety. Virtual tests, where autonomous vehicles are tested in software simulation, are common practices because they are more efficient and safer compared to field operational tests. Specifically, search-based approaches are used to find particularly critical situations. These approaches provide an opportunity to automatically generate tests; however, systematically producing bug-revealing tests for ADS remains a major challenge. To address this challenge, we introduce DoppelTest, a test generation approach for ADSes that utilizes a genetic algorithm to discover bug-revealing violations by generating scenarios with multiple autonomous vehicles that account for traffic control (e.g., traffic signals and stop signs). Our extensive evaluation shows that DoppelTest can efficiently discover 123 bug-revealing violations for a production-grade ADS (Baidu Apollo) which we then classify into 8 unique bug categories.