Mô hình hóa Resilient Modulus đất bằng Regression và Neural Network

UNIVERSITY OF OKLAHOMA GRADUATE COLLEGE REGRESSION AND NEURAL NETWORK MODELING OF RESILIENT MODULUS BASED ON ROUTINE SOIL PROPERTIES AND STRESS STATES A DISSERTATION SUBMITTED TO THE GRADUATE FACULTY in partial fulfillment of the requirements for the Degree of Doctor of Philosophy By ALI HONARMAND-EBRAHIMI Norman, Oklahoma 2006 UMI Number: 3238559 UMI Microform 3238559 Copyright 2007 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.

Box 1346 Ann Arbor, MI 48106-1346 © Copyright by ALI HONARMAND-EBRAHIMI 2006 All Rights Reserved. ACKNOWLEDGMENTS I would like to express my deepest gratitude to my committee chairman, Dr. Musharraf Zaman for his patience, understanding, and believing in me for the past twenty-two years. This work would not have been completed without Dr.

My sincere appreciation and thanks goes to Dr. Joakim Laguros, Dr. Gerald Miller, Dr. Kianoosh Hatami and Dr.

Luther White for serving in my committee and providing me with the information I needed during my dissertation work. I would like to thank Mr. Chau Yih Cheong for his assistance in completing this work. My sincere appreciation also goes to the staff and engineers of Burgess Engineering and Testing, Inc., especially to Mr.

Hai Ming Lim and Mr. Cheong Fung Ting for their assistance. Finally, I would like to thank my wife Ronessa for standing by me no matter how rough the sea and the journey has been. Thank you for all your support and the two beautiful children, Esther and Ali.

My appreciation also goes to my parents, Issa and Mehri, for their help and support. iv TABLE OF CONTENTS ACKNOWLEDGMENTS .iv TABLE OF CONTENTS .v LIST OF TABLES.ix LIST OF FIGURES .xxi CHAPTER 1 INTRODUCTION .2 Objectives and Study Tasks.3 Format of the Dissertation .9 CHAPTER 2 LITERATURE REVIEW .3 Determination of Resilient Modulus From Laboratory and In-situ Testing.2 In-situ Test .3 Comparison of Laboratory and In-situ Test Results.4 Variability in Resilient Modulus Testing .5 Influence of Resilient Modulus on Pavement Performance .6 Determination of Resilient Moduli from Correlations with Other Soil Properties .7 Types of Generalized Linear Model.8 Artificial Neural Network.9 Artificial Neural Network Model .35 CHAPTER 3 MATERIAL SOURCES AND EXPERIMENTAL METHODOLOGY .4 Soil Classification Tests .7 Resilient Modulus Test.1 Testing Equipment and Setup.8 Unconfined Compression Test .55 CHAPTER 4 PRESENTATION OF RESULTS .1 Material parameters and their Relationship with M-E Pavement Design .5 Grain Size Distribution .10 Unconfined Compressive Strength .3 Resilient Modulus Test.86 CHAPTER 5 STATISTICAL MODELS .2 Application of Existing Models.3 Overview of Statistical Models .1 Stress-Based Model .2 Multiple Regression Model .1 Stress-Based Model Development.2 Multiple Regression Model Development.3 Polynomial Model Development.4 Factorial Model Development .5 Comments On Comparative Performance of the Statistical Models .5 Evaluation of Models.1 Evaluation of Factorial Model .2 Evaluation of second order Polynomial Model .126 CHAPTER 6 ARTIFICIAL NEURAL NETWORK .2 Artificial Neural Network Models.2 General Regression Neural Network (GRNN) .3 Radial Basis Function Network (RBFN).4 Multi-Layer Perceptrons Network (MLPN) .1 Linear Network (LN) Development .2 General Regression Neural Network (GRNN) Development .3 Radial Basis Function Network (RBFN) Development .4 Multi-Layer Perceptrons Network (MLPN) Development.5 Comments On Comparative Performance of the ANN Models .4 Evaluation of Models.1 Evaluation of LN Model.2 Evaluation of GRNN Model.3 Evaluation of RBFN Model.4 Evaluation of MLPN Models.1 Sensitivity Analysis For LN Model.2 Sensitivity Analysis For GRNN Model.3 Sensitivity Analysis For RBFN Model.4 Sensitivity Analysis For MLPN Models.6 Alternative MLPN-2 Models.7 Design Chart for Application of MLPN-2 Model .172 CHAPTER 7 PAVEMENT DESIGN APPLICATION .2 Resilient Modulus for Pavement Design .4 Pavement Design Results .242 CHAPTER 8 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS .255 APPENDIX A Development and Evaluation Datasets .266 APPENDIX B Resilient Modulus Test Results .276 APPENDIX C Equation for Factorial Model.285 APPENDIX D Predictions of Resilient Modulus from Statistical and Artificial Neural Network Models .287 viii LIST OF TABLES Table 2-1 Comparison of Different AASHTO Test Methods for Resilient Modulus Testing of Granular Base/Subbase Materials .38 Table 2-2 Comparison of AASHTO Test Methods for Resilient Modulus Testing of Base/Subbase and Subgrade Materials.39 Table 2-3 Summary of Correlation Equations from Literature for Resilient Modulus .40 Table 3-1 Soil Series and Parent Materials for the Sample Used in Model Development.57 Table 3-2 Soil Series and Parent Materials for the Sample Used in Model Evaluation.59 Table 3-3 Distribution of Parent Materials in Experiment and Evaluation Datasets.60 Table 3-4 Resilient Modulus Testing Sequence for Subgrade Soil (AASHTO, 2004).60 Table 3-5 Resilient Modulus Testing Results for Sample AL-8A.61 Table 4-1 Basic Statistical Parameters for Liquid Limit (LL) .88 Table 4-2 Basic Statistical Parameters for Plastic Limit (PL) .88 Table 4-3 Basic Statistical Parameters for Plasticity Index (PI).89 Table 4-4 Basic Statistical Parameters for Percent Passing 4.89 Table 4-5 Basic Statistical Parameters for Percent Passing 2.90 Table 4-6 Basic Statistical Parameters for Percent Passing 0.90 Table 4-7 Basic Statistical Parameters for Percent Passing 0.91 Table 4-8 Summary of Soil Classification Results for the Development Dataset .92 Table 4-9 Summary of Soil Classification Results for the Evaluation Dataset .92 ix Table 4-10 Basic Statistical Parameters for Group Index (GI).93 Table 4-11 Basic Statistical Parameters for Specimen Moisture Content .93 Table 4-12 Basic Statistical Parameters for Specimen Dry Density.94 Table 4-13 Basic Statistical Parameters for Unconfined Compressive Strength.94 Table 4-14 Basic Statistical Parameters for Resilient Modulus at Each Sequence for Development Dataset (126 Specimens).95 Table 4-15 Basic Statistical Parameters for Resilient Modulus at Each Sequence for Evaluation Dataset (68 Specimens).96 Table 4-16 Basic Statistical Parameters for Resilient Modulus at Each Sequence for Evaluation Dataset at Rogers County (58 Specimens).97 Table 4-17 Basic Statistical Parameters for Resilient Modulus at Each Sequence for Evaluation Dataset at Woodward County (10 Specimens) .98 Table 5-1 Summary of R2 and F Values for the Statistical Modeling .128 Table 5-2 Summary of the Statistical Modeling Results .129 Table 6-1 Summary of the Artificial Neural Network (ANN) Modeling Results .173 Table 6-2 Average and Standard Deviation of the Independent Variables from Development and Evaluation Datasets .173 Table 6-3 Sensitivity Study for the Linear Network (LN) Model .174 Table 6-4 Sensitivity Study for the Generalized Regression Neural Network (GRNN) Model.175 Table 6-5 Sensitivity Study for the Radial Basis Function Network (RBFN) Model.176 Table 6-6 Sensitivity Study for the Multilayer Perceptrons Network (MLPN-1) Model.177 Table 6-7 Sensitivity Study for the Multilayer Perceptrons Network (MLPN-2) Model.178 Table 6-8 Sensitivity Study for the Multilayer Perceptrons Network (MLPN-3) Model.179 x Table 6-9 Sensitivity Study for the Multilayer Perceptrons Network (MLPN- 3) Model .180 Table 6-10 Ranking of Independent Variables for ANN Models.181 Table 6-11 Ranking of Independent Variables for ANN Modeling Results for Separated Specimens Based on Moisture.181 Table 6-12 Percent Data Occurring in Each Zone Corresponding to the Range of MR/Pa .182 Table 7-1 Subgrade Resilient Moduli for Pavement Design .244 Table 7-2 Equivalent Single Axle Load (ESAL) Calculation.244 Table 7-3 Reliability and Serviceability .245 Table 7-4 Pavement Design Results .245 Table 7-5 Comparison of the Effect of Reliability in SN and Pavement Design .245 xi LIST OF FIGURES Figure 2-1 Typical Stress-Strain Response from Repeated Load Test (After Elliott and Thornton, 1988) .44 Figure 2-2 Range of k1 and k2 Values for Aggregates (After Rada and Witczak, 1981) .45 Figure 2-3 Typical Bi-linear Model .46 Figure 2-4 A General Feedforward Nueral Network Architecture (StatSoft, Inc.47 Figure 2-5 A Typical Processing Element (StatSoft, Inc.48 Figure 3-1 Flow Chart of Operations .64 Figure 3-2 Location Map for Soil Samples Used in Laboratory Testing and Model Development .65 Figure 3-3 Location Map for Additional Soils Samples Used in Model Evaluation.66 Figure 3-4 Moisture-Density Relationship Curve for Sample AL-10 .67 Figure 3-5 Static Compaction Process (AASHTO, 2004) .68 Figure 3-6 Resilient Modulus Testing System.69 Figure 3-7 Detail Illustration of Electronics Bay and Pump Bay .70 Figure 3-8 Triaxial Chamber Setup for Resilient Modulus Test.71 Figure 3-9 Haversine-shaped Load Pulse in a Loading Cycle .72 Figure 3-10 Unconfined Compression Test Results for Sample AL-8A .73 Figure 4-1 Distribution of Liquid Limits (LL) for the Development and the Evaluation Datasets .99 Figure 4-2 Distribution of Liquid Limits (LL) for the Evaluation Dataset for Rogers County and Woodward County Soils.99 Figure 4-3 Distribution of Plastic Limits (PL) for the Development and the Evaluation Datasets .100 Figure 4-4 Distribution of Plastic Limits (PL) for the Evaluation Dataset for Rogers County and Woodward County Soils.100 xii Figure 4-5 Distribution of Plasticity Index (PI) for the Development and the Evaluation Datasets .101 Figure 4-6 Distribution of Plasticity Index (PI) for the Evaluation Dataset for Rogers County and Woodward County Soils.101 Figure 4-7 Distribution of Percent Passing 4. 4) Sieve for the Development and the Evaluation Datasets .102 Figure 4-8 Distribution of Percent Passing 4. 4) Sieve for the Evaluation Dataset for Rogers County and Woodward County Soils .102 Figure 4-9 Distribution of Percent Passing 2.

10) Sieve for the Development and the Evaluation Datasets .103 Figure 4-10 Distribution of Percent Passing 2. 10) Sieve for the Evaluation Dataset for Rogers County and Woodward County Soils .103 Figure 4-11 Distribution of Percent Passing 0. 40) Sieve for the Development and the Evaluation Datasets .104 Figure 4-12 Distribution of Percent Passing 0. 40) Sieve for the Evaluation Dataset for Rogers County and Woodward County Soils .104 Figure 4-13 Distribution of Percent Passing 0.

200) Sieve for the Development and the Evaluation Datasets .105 Figure 4-14 Distribution of Percent Passing 0. 200) Sieve for the Evaluation Dataset for Rogers County and Woodward County Soils .105 Figure 4-15 Distribution of USCS Soil Classification for the Development and the Evaluation Datasets .106 Figure 4-16 Distribution of USCS Soil Classification for the Evaluation Dataset for Rogers County and Woodward County Soils .106 Figure 4-17 Distribution of AASHTO Soil Classification for the Development and the Evaluation Datasets .107 xiii Figure 4-18 Distribution of AASHTO Soil Classification for the Evaluation Dataset for Rogers County and Woodward County Soils .107 Figure 4-19 Distribution of Group Index (GI) for the Development and the Evaluation Datasets .108 Figure 4-20 Distribution of Group Index (GI) for the Evaluation Dataset for Rogers County and Woodward County Soils.108 Figure 4-21 Distribution of Specimen Moisture Content for the Development and the Evaluation Datasets .109 Figure 4-22 Distribution of Specimen Moisture Content for the Evaluation Dataset in Rogers County and Woodward County Soils .109 Figure 4-23 Distribution of Specimen Dry Density for the Development and the Evaluation Datasets .110 Figure 4-24 Distribution of Specimen Dry Density for the Evaluation Dataset for Rogers County and Woodward County Soils .110 Figure 4-25 Distribution of Unconfined Compressive Strength for the Development and the Evaluation Datasets .111 Figure 4-26 Distribution of Unconfined Compressive for the Evaluation Dataset for Rogers County and Woodward County Soils .111 Figure 5-1 Comparison of Experimental and Predicted MR/Pa for Development Dataset: Moossazadeh and Witczak, 1981 Stress- Based Model.130 Figure 5-2 Comparison of Experimental and Predicted MR/Pa for Development Dataset: NCHRP, 2003 Stress-Based Model .131 Figure 5-3 Comparison of Experimental and Predicted MR/Pa for Development Dataset: NCHRP, 2003 Stress-Based Model for One Confining Pressure .132 Figure 5-4 Comparison of Experimental and Predicted MR/Pa for Development Dataset: NCHRP, 2004 Stress-Based Model for one MR Test (Three Confining Pressures.133 xiv Figure 5-5 Comparison of Experimental and Predicted MR/Pa for Development Dataset: Stress-Based Model .134 Figure 5-6 Comparison of Experimental and Predicted MR/Pa for Development Dataset: Multiple Regression Model .135 Figure 5-7 Resilient Modulus from Experiment and Multiple Regression Model: Specimen MA-3B .136 Figure 5-8 Resilient Modulus from Experiment and Multiple Regression Model: Specimen NO-7A.137 Figure 5-9 Resilient Modulus from Experiment and Multiple Regression Model: Specimen OS-1B.138 Figure 5-10 Comparison of Experimental and Predicted MR/Pa for Development Dataset: Polynomial Model .139 Figure 5-11 Resilient Modulus from Experiment and Polynomial Model: Specimen MA-3B.140 Figure 5-12 Resilient Modulus from Experiment and Polynomial Model: Specimen NO-7A .141 Figure 5-13 Resilient Modulus from Experiment and Polynomial Model: Specimen OS-1B .142 Figure 5-14 Comparison of Experimental and Predicted MR/Pa for Development Dataset: Factorial Model.143 Figure 5-15 Resilient Modulus from Experiment and Factorial Model: Specimen MA-3B.