Artificial Neural Networks in Vehicular Pollution Modelling Book

Introduction     1
Air Pollution Definition     2
Composition of Atmosphere     2
Air Pollution Problems     3
Air Pollution Sources     4
Point Source Emissions     4
Area Source Emissions     4
Line Source Emissions     5
Urban Air Pollution Control Strategies     5
Modelling Tools - Conventional and Soft Computational Approach Including ANN     5
Vehicular Pollution     7
General     7
Sources of Vehicular Pollution     8
Types of Vehicular Pollutants     10
Carbon Monoxide     10
Nitrogen Oxides     10
Volatile Organic Compounds     11
Sulphur Dioxide     11
Particulate Matter     12
Lead     12
Health Effects of Vehicular Pollution     13
Meteorological and Topographical Factors Affecting Vehicular Pollution Dispersion in Urban Air Sheds     15
Ambient Air Quality Monitoring     18
Local Air Quality Management     19
Options for Control of Vehicular Pollution     22
Ambient Air Quality Standards     23
Overview of Vehicular Pollution Modelling     23
Artificial Neutral Networks     25
General     25
What Artificial Neural Networks are?     25
Basic Concepts of Neural Network     26
Human Biological Neuron     26
Simple Neuron Model     28
History of Artificial Neural Network     29
Artificial Neural Network Architecture     30
Types of Neural Networks     31
Feed-Forward Networks     32
Recurrent Neural Networks     32
Transfer Functions and Learning Algorithms     34
Transfer Functions     34
Learning Methods     34
Back-Propagation Learning Algorithm     35
Summary     39
Vehicular Pollution Modelling-Conventional Approach     41
General     41
Theoretical Approaches of Vehicular Pollution Modelling     42
Vehicular Pollution Deterministic Models     47
Vehicular Pollution Numerical Models     55
Vehicular Pollution Stochastic Models     58
ANN based Vehicular Pollution Models     61
Limitations of Vehicular Pollution Models     63
Summary     66
Vehicular Pollution Modelling - ANN Approach     67
General      67
ANN Approach to Vehicular Pollution Modelling     68
Algorithm for ANN based Vehicular Pollution Model     69
Selection of the Optimal ANN based Vehicular Pollution Model Architecture     70
Selection of the Best Activation Functions     71
Selection of the Optimum Learning Parameters     71
Initialization of the Network Weights and Bias     72
Training Procedure     73
Statistics for Testing ANN based Vehicular Pollution Models     77
Development of ANN based Vehicular Pollution Models     78
Case Study     79
Pollutant Data     81
Traffic Data     84
Meteorological Data     85
Models Development     86
Summary     119
Application of ANN based Vehicular Pollution Models     121
General     121
Model Performance Indicators     122
Root Mean Square Error     122
Coefficient of Determination     123
Mean Bias Error     124
Standard Deviations     124
Slope and Intercept of the Least Square Regression Equation     125
Degree of Agreement     125
Application of ANN Based Vehicular Pollution Models at Urban Intersection and Straight Road Corridor     125
1-hr Average CO Models     125
8-hr Average CO Models     133
24-hr Average NO[subscript 2] Models     140
Performance Evaluation and Comparison of ANN based Vehicular Pollution Models with Conventional Models     147
Performance of ANN based CO Models for the Critical Period Test Data     147
Performance of Univariate Stochastic Models for the Critical Period Test Data     149
Performance of Deterministic Model for the Critical Period Test Data     151
Summary     155
Epilogue     157
Appendix A     163
Appendix B     175
Appendix C     185
Appendix D     211
References     227