A robust prediction model for evaluation of plastic limit based on sieve # 200 passing material using gene expression...

Area of Science:

• Geotechnical Engineering
• Soil Mechanics
• Artificial Intelligence in Civil Engineering

Background:

• The plastic limit (PL) is crucial for classifying fine-grained soils, typically determined using material passing sieve #40.
• Accurate PL determination should ideally use material passing sieve #200 (PL200), but laboratory methods are time-consuming and complex.
• PL200 is influenced by soil composition (sand, silt, clay) and conventional methods are inadequate.

Purpose of the Study:

• To develop a novel, high-accuracy prediction model for the plastic limit of soil using sieve #200 passing material (PL200).
• To utilize Gene Expression Programming (GEP), an artificial intelligence technique, for modeling PL200.
• To establish a reliable and efficient method for predicting PL200, overcoming limitations of traditional laboratory testing.

Main Methods:

• Collected laboratory experimental data on soil properties.
• Developed a prediction model for PL200 using Gene Expression Programming (GEP).
• Input parameters included sand, clay, silt content, and plastic limit using sieve #40 material (PL40).
• Validated the model using statistical metrics (R2, RMSE, MAE, RSE) and performed sensitivity/parametric analyses.

Main Results:

• The GEP model demonstrated high accuracy in predicting PL200.
• Statistical validation confirmed the model's reliability and performance.
• Sensitivity and parametric studies supported the accuracy and robustness of the proposed prediction model.

Conclusions:

• The developed GEP model provides an accurate and reliable method for predicting PL200.
• The model offers a viable and efficient alternative to traditional, time-consuming laboratory PL determination.
• The proposed AI-based approach can be effectively utilized in field applications for soil classification.