Scholarlink

<p>The California Bearing Ratio (CBR) test is a penetration test used to evaluate the strength of subgrade soils for pavement design. The laboratory procedure of determining the CBR of pavement material is lengthy and time-consuming. The present study introduces an optimal performance model by comparing genetic&#xa0;algorithm (GA) and particle swarm (PSO)-optimized least square support vector machine (LSSVM), relevance vector machine (RVM), and artificial neural network (ANN) to predict the CBR of hydrated lime-activated rice husk ash (HARHA)-treated soil. These models were developed and analyzed with 80% (= 97) training and 20% (= 24) testing datasets. The consistency and compaction parameters, clay activity, and HARHA content have been utilized as features. The performance comparison revealed that the GA_RVM model outperformed GA_LSSVM, PSO_LSSVM, PSO_RVM, GA_ANN, and PSO_ANN models with a performance index of 1.99, root mean square error&#xa0;of 0.3158% and variance accounted for of 99.92. The score, regression error characteristic, and accuracy matrix analyzed the prediction capabilities and reliability of the GA_RVM model. In addition, the uncertainty and generalizability determined the robustness of the GA_RVM model in predicting the CBR. The cosine amplitude sensitivity analysis showed that the HARHA content (= 99.54%) is the most significant feature in predicting CBR, followed by maximum dry density (= 95.02%) and plastic limit (= 83.07%).</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

Prediction of CBR of Hydrated Lime-Activated Rice Husk Ash (HARHA) Treated Soil Using Genetic and Particle Swarm Optimized ANN, LSSVM, and RVM Models

  • 期刊论文

  • 发表日期：

    2025年7月9日

查看全文
  • Subodh Kumar Suman,
  • Shashank Shekhar Kamal,
  • Avinash Kumar,
  • Jitendra Khatti,
  • Abidhan Bardhan

摘要

The California Bearing Ratio (CBR) test is a penetration test used to evaluate the strength of subgrade soils for pavement design. The laboratory procedure of determining the CBR of pavement material is lengthy and time-consuming. The present study introduces an optimal performance model by comparing genetic algorithm (GA) and particle swarm (PSO)-optimized least square support vector machine (LSSVM), relevance vector machine (RVM), and artificial neural network (ANN) to predict the CBR of hydrated lime-activated rice husk ash (HARHA)-treated soil. These models were developed and analyzed with 80% (= 97) training and 20% (= 24) testing datasets. The consistency and compaction parameters, clay activity, and HARHA content have been utilized as features. The performance comparison revealed that the GA_RVM model outperformed GA_LSSVM, PSO_LSSVM, PSO_RVM, GA_ANN, and PSO_ANN models with a performance index of 1.99, root mean square error of 0.3158% and variance accounted for of 99.92. The score, regression error characteristic, and accuracy matrix analyzed the prediction capabilities and reliability of the GA_RVM model. In addition, the uncertainty and generalizability determined the robustness of the GA_RVM model in predicting the CBR. The cosine amplitude sensitivity analysis showed that the HARHA content (= 99.54%) is the most significant feature in predicting CBR, followed by maximum dry density (= 95.02%) and plastic limit (= 83.07%).