Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Hybrid Metaheuristic Optimization of Artificial Neural Networks for Liquefaction Probability Prediction Using Various Historical CPT Data
https://orcid.org/http://orcid.org/0000-0003-1777-9981
This work aims to forecast soil liquefaction probabilities by adopting artificial neural networks (ANNs) optimized by various algorithms. The optimization algorithms applied included teaching–learning-based optimization (ANN-TLBO), imperialist competitive algorithm (ANN-ICA), and shuffled complex evolution (ANN-SCE). Additionally, ant colony optimization (ANN-ACO), ant lion optimizer (ANN-ALO), and artificial bee colony (ANN-ABC) were also utilized. A reliability analysis, COM value, a new a20-index, accuracy matrix, tailor diagram, and sensitive analysis were performed to compare the accuracy of the prediction models. The results showed that the ANN-TLBO model achieved the highest performance, with a score of 94 and a COM value of 0.1477. It outperformed the other models, including ANN-ALO (73 and 0.1822), ANN-ICA (56 and 0.1935), ANN-ABC (55 and 0.2257), ANN-ACO (44 and 0.2322), and ANN-SCE (28 and 0.2327). The statistical results highlight the ANN-TLBO model’s superior precision (R2 = 0.708 for training and R2 = 0.750 for the testing) and minimal error (RMSE = 0.163 for training and RMSE = 0.158 for the testing). Additionally, external validation and comparative analysis were employed to assess the robustness of the models. This study underscores the effectiveness of combining ANN with advanced optimization techniques to enhance the prediction of soil liquefaction.
Hybrid Metaheuristic Optimization of Artificial Neural Networks for Liquefaction Probability Prediction Using Various Historical CPT Data
https://orcid.org/http://orcid.org/0000-0003-1777-9981
This work aims to forecast soil liquefaction probabilities by adopting artificial neural networks (ANNs) optimized by various algorithms. The optimization algorithms applied included teaching–learning-based optimization (ANN-TLBO), imperialist competitive algorithm (ANN-ICA), and shuffled complex evolution (ANN-SCE). Additionally, ant colony optimization (ANN-ACO), ant lion optimizer (ANN-ALO), and artificial bee colony (ANN-ABC) were also utilized. A reliability analysis, COM value, a new a20-index, accuracy matrix, tailor diagram, and sensitive analysis were performed to compare the accuracy of the prediction models. The results showed that the ANN-TLBO model achieved the highest performance, with a score of 94 and a COM value of 0.1477. It outperformed the other models, including ANN-ALO (73 and 0.1822), ANN-ICA (56 and 0.1935), ANN-ABC (55 and 0.2257), ANN-ACO (44 and 0.2322), and ANN-SCE (28 and 0.2327). The statistical results highlight the ANN-TLBO model’s superior precision (R2 = 0.708 for training and R2 = 0.750 for the testing) and minimal error (RMSE = 0.163 for training and RMSE = 0.158 for the testing). Additionally, external validation and comparative analysis were employed to assess the robustness of the models. This study underscores the effectiveness of combining ANN with advanced optimization techniques to enhance the prediction of soil liquefaction.
Hybrid Metaheuristic Optimization of Artificial Neural Networks for Liquefaction Probability Prediction Using Various Historical CPT Data
https://orcid.org/http://orcid.org/0000-0003-1777-9981
This work aims to forecast soil liquefaction probabilities by adopting artificial neural networks (ANNs) optimized by various algorithms. The optimization algorithms applied included teaching–learning-based optimization (ANN-TLBO), imperialist competitive algorithm (ANN-ICA), and shuffled complex evolution (ANN-SCE). Additionally, ant colony optimization (ANN-ACO), ant lion optimizer (ANN-ALO), and artificial bee colony (ANN-ABC) were also utilized. A reliability analysis, COM value, a new a20-index, accuracy matrix, tailor diagram, and sensitive analysis were performed to compare the accuracy of the prediction models. The results showed that the ANN-TLBO model achieved the highest performance, with a score of 94 and a COM value of 0.1477. It outperformed the other models, including ANN-ALO (73 and 0.1822), ANN-ICA (56 and 0.1935), ANN-ABC (55 and 0.2257), ANN-ACO (44 and 0.2322), and ANN-SCE (28 and 0.2327). The statistical results highlight the ANN-TLBO model’s superior precision (R2 = 0.708 for training and R2 = 0.750 for the testing) and minimal error (RMSE = 0.163 for training and RMSE = 0.158 for the testing). Additionally, external validation and comparative analysis were employed to assess the robustness of the models. This study underscores the effectiveness of combining ANN with advanced optimization techniques to enhance the prediction of soil liquefaction.
Hybrid Metaheuristic Optimization of Artificial Neural Networks for Liquefaction Probability Prediction Using Various Historical CPT Data
Transp. Infrastruct. Geotech.
B., Dhilipkumar (Autor:in) / Samui, Pijush (Autor:in)
01.01.2025
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Roughness Level Probability Prediction Using Artificial Neural Networks
| British Library Online Contents | 1997
Roughness Level Probability Prediction Using Artificial Neural Networks
| British Library Conference Proceedings | 1997
Prediction of Liquefaction Susceptibility of Subsoil Layers Using Artificial Neural Networks
| Springer Verlag | 2024