Hybrid neuro-fuzzy ML and MC simulation-based Reliability Analysis of Simply Supported Beam

KUMAR, ABHISHEK; Rai, Baboo; Samui, Pijush

doi:10.22059/ceij.2024.374883.2045

Hybrid neuro-fuzzy ML and MC simulation-based Reliability Analysis of Simply Supported Beam

Articles in Press

Authors

¹ RESEARCH SCHOLAR, DEPARTMENT OF CIVIL ENGINEERING, NATIONAL INSTITUTE OF TECHNOLOGY, PATNA, INDIA

² Department of civil engineering, Associate Professor, National Institute of Technology, Patna, India

³ Department of Civil Engineering, NIT Patna, Patna, Bihar, India

10.22059/ceij.2024.374883.2045

Abstract

The paper introduces an innovative approach utilizing hybrid Adaptive Neuro Fuzzy Inference System (ANFIS) models for the reliability-based design of structural beams. While structural reliability analysis with hybrid ANFIS models remains largely unexplored, existing studies primarily rely on rudimentary simulation models. To address this gap, our study employs Particle Swarm Optimization (PSO) and Genetic Algorithms (GA) to enhance the performance of the ANFIS model. The ML models are validated on three Monte-Carlo datasets of size 1000, 2500 and 5000 respectively. The study findings demonstrate satisfactory performance across all ML models, with the hybrid ANFIS models exhibiting superior predictive capabilities compared to traditional methods. Among the hybrid ANFIS models, ANFIS-PSO emerges as the most robust. The reliability indices and Probability of Failure (POF) values are calculated for the predicted values and compared with actual values. It is concluded that ANFIS-PSO based methodology is the most robust model and outperforms the other models. It is noteworthy that while the ANFIS-PSO model demonstrates exceptional performance, all models presented in this study serve as valuable tools for reliability-based structural design, offering robust alternatives to conventional methodologies.

Keywords