Scholarlink

<p>This study investigates the suitability of Monte Carlo simulation (MCS) for reliability analysis (RA) of a back-to-back reinforced soil (RS) wall considering external stability. Furthermore, hybrid artificial neural network (ANN) frameworks based on the first-order second-moment method (FOSM) were applied to automate the process of RA. Six hybrid ANNs were built utilizing aquila optimizer, colony predation algorithm, firefly algorithm, grey wolf optimizer, marine predators algorithm, and particle swarm optimization. The RA was performed at various loading combinations and co-efficient of variation (COV) levels. The efficiency of the hybrid ANNs was assessed using several performance indices, and the best-fitted hybrid ANN was then used to automate the process of RA against sliding, overturning, and bearing failures. In the testing phase, the hybrid framework of ANN and firefly algorithm, ANN-FF, provides the best-fitted estimation of FOS against sliding, overturning, and bearing, with correlation coefficients of 99.99%, 99.99%, and 99.75%, respectively. Initially, the probability of failure was determined using the MCS followed by utilization of the ANN-FF framework to automate the process of RA based on MCS-based samples. Overall, the FOSM-based ANN-FF framework can be considered an alternate approach for risk assessment of back-to-back RS walls under varying COV levels.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reliability Analysis of External Stabilities of Reinforced Soil Wall Using Monte Carlo Simulation and an Efficient Hybrid Artificial Neural Network Paradigm

  • 期刊论文

  • 发表日期：

    2025年1月17日

查看全文
  • Sudeep Kumar,
  • Avijit Burman

摘要

This study investigates the suitability of Monte Carlo simulation (MCS) for reliability analysis (RA) of a back-to-back reinforced soil (RS) wall considering external stability. Furthermore, hybrid artificial neural network (ANN) frameworks based on the first-order second-moment method (FOSM) were applied to automate the process of RA. Six hybrid ANNs were built utilizing aquila optimizer, colony predation algorithm, firefly algorithm, grey wolf optimizer, marine predators algorithm, and particle swarm optimization. The RA was performed at various loading combinations and co-efficient of variation (COV) levels. The efficiency of the hybrid ANNs was assessed using several performance indices, and the best-fitted hybrid ANN was then used to automate the process of RA against sliding, overturning, and bearing failures. In the testing phase, the hybrid framework of ANN and firefly algorithm, ANN-FF, provides the best-fitted estimation of FOS against sliding, overturning, and bearing, with correlation coefficients of 99.99%, 99.99%, and 99.75%, respectively. Initially, the probability of failure was determined using the MCS followed by utilization of the ANN-FF framework to automate the process of RA based on MCS-based samples. Overall, the FOSM-based ANN-FF framework can be considered an alternate approach for risk assessment of back-to-back RS walls under varying COV levels.