Optimized kernel extreme learning machine using Sine Cosine Algorithm for prediction of unconfined compression strength of MICP cemented soil

• Published in: Environmental science and pollution research international Vol. 31; no. 17; pp. 24868 - 24880
• Main Authors: Peng, Shuquan, Sun, Qiangzhi, Fan, Ling, Zhou, Jian, Zhuo, Xiande
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.04.2024; Springer Nature B.V
• Subjects: Algorithms; Aquatic Pollution; Artificial neural networks; Atmospheric Protection/Air Quality Control/Air Pollution; Bioremediation; Calcite; Chemical precipitation; compression strength; Compressive strength; data collection; Earth and Environmental Science; Ecotoxicology; Environment; Environmental Chemistry; Environmental Health; Kernel functions; Learning algorithms; Machine learning; Optimization; Performance evaluation; Population number; population size; prediction; Regularization; Research Article; soil; Soil remediation; Soils; Support vector machines; Trigonometric functions; Waste Water Technology; Water Management; Water Pollution Control; SCA; KELM; Sensitivity; UCS; MICP
• ISSN: 1614-7499; 0944-1344; 1614-7499
• DOI: 10.1007/s11356-024-32687-2

Microbially induced calcite precipitation (MICP) is an eco-friendly bio-remediation technology. The unconfined compressive strength (UCS) of MICP cemented soil is an important indicator of repair effectiveness. This study proposes a machine learning technique utilizing the Sine Cosine Algorithm (SCA) to optimize the regularization coefficient C and kernel width γ of the kernel extreme learning machine (KELM) to predict the UCS of MICP cemented soil. To evaluate the performance of the proposed models, a dataset containing 180 groups of the UCS of MICP cemented soil was obtained. The results obtained by SCA-KELM were compared with those obtained by the Random Forest algorithm (RF), Support Vector Machine (SVM), and KELM. The performance of these models was evaluated by the scores of MAE, RMSE, and R 2 . The results indicate that the SCA-KELM algorithm exhibits optimal prediction performance (Total score: 21). After optimizing KELM with SCA, the total score improved by 110%, suggesting that SCA significantly enhances the KELM performance. After model development, the optimal population size for SCA-KELM was determined to be 50. Based on the mutual information test, an innovative method was developed for categorizing factor sensitivity by employing importance scores as the partitioning criterion. This method categorizes the influencing factors into three tiers: high (importance score: 8.03–11.14%), medium (importance score: 5.93–7.25%), and low (importance score: 3.23–5.18%). These results suggest that the proposed SCA-KELM algorithm can be regarded as a powerful tool for predicting the UCS of MICP cemented soil.