Soft Computing in Water Resources Engineering: Artifical Neural Networks, Fuzzy Logic and Genetic Algorithms

Engineers have attempted to solve water resources engineering problems with the help of empirical, regression-based and numerical models. Empirical models are not universal, nor are regression-based models. The numerical models are, on the other hand, physics-based but require substantial data measurement and parameter estimation. Hence, there is a need to employ models that are robust, user-friendly, and practical and that do not have the shortcomings of the existing methods. Artificial intelligence methods meet this need. Soft Computing in Water Resources Engineering introduces the basics of artificial neural networks (ANN), fuzzy logic (FL) and genetic algorithms (GA). It gives details on the feed forward back propagation algorithm and also introduces neuro-fuzzy modelling to readers. Artificial intelligence method applications covered in the book include predicting and forecasting floods, predicting suspended sediment, predicting event-based flow hydrographs and sedimentographs, locating seepage path in an earth-fill dam body, and the predicting dispersion coefficient in natural channels.The author also provides an analysis comparing the artificial intelligence models and contemporary non-artificial intelligence methods (empirical, numerical, regression, etc.). The ANN, FL, and GA are fairly new methods in water resources engineering. The first publications appeared in the early 1990s and quite a few studies followed in the early 2000s. Although these methods are currently widely known in journal publications, they are still very new for many scientific readers and they are totally new for students, especially undergraduates. Numerical methods were first taught at the graduate level but are now taught at the undergraduate level. There are already a few graduate courses developed on AI methods in engineering and included in the graduate curriculum of some universities. It is expected that these courses, too, will soon be taught at the undergraduate levels.

Table of Contents:
PART I - ARTIFICIAL NEURAL NETWORKS Chapter 1 - Introduction to Artificial Neural Networks General View; Biological Neuron; Artificial Neuron; Artificial Neural Network; History; General Properties of ANN; Types; Architecture; Neuro-dynamics; ANN versus Other Models Chapter 2 - Artificial Neuron Components of Artificial Neuron; Methods for Computing Net Information; Summation (P) method; Maximum (max) method; Minimum (min) method; Product (Q) method; Activation Functions; Linear function; Step function; Rampage function; Gaussian function; Sigmoid function; Hyperbolic tangent function Chapter 3 - Network Training Pre-Training Procedures; Data Standardization; Standardization methods when using sigmoid function; Standardization methods when using hyperbolic tangent function; Network Initialization; Network Training; Back-propagation algorithm; Updating weights in output-inner layers; Updating weights in inner-input layers; Worked examples; Radial basis function; Conjugate gradient algorithm; Cascade correlation algorithm; Generalized regression algorithm; Learning Rules; Learning Parameter; Appendix; Exercise Problem Chapter 4 - Model Testing De-standardization of Model Output; Evaluating Model Performance; Over-training and Cross-training Chapter 5 - Model Application in Water Resources Engineering Prediction; Total suspended sediment; Seepage; Dispersion coefficient; Sheet sediment; Runoff at plot scale; Runoff at watershed scale; Flood hydrograph at basin scale; Classification; Forecasting; Forecasting flood hydrograph at basin scale; Extrapolation; Filling Gap in Time Series Data; References PART II - FUZZY LOGIC ALGORITHM Chapter 6 - Introduction to Fuzzy Logic Algorithm General View; Basic Concept in Fuzzy Logic; Fuzzy Systems Chapter 7 - Fuzzy Membership Functions, Set Operations, and Fuzzy Relations Fuzzy Membership Functions; Fuzzy Set Operations; Set representation; Set operations; Union of sets; Intersection of sets; Complementary sets; Subsets; Operation properties of fuzzy sets; Operations unique to fuzzy sets; Concentration; Dilation; Normalization; Intensification; Fuzzy Relations; Exercise Questions Chapter 8 - Constructing Fuzzy Model Fuzzification; Fuzzy Rule Base;Fuzzy Inference Engine; Inference sub-process; Composition sub-process; Defuzzification; Exercise Questions Chapter 9 - Fuzzy Model Application in Water Resources Engineering Introduction; TSS Prediction; Model development; Model calibration and application; Sheet Sediment Prediction; Fuzzy model; Physics-based model; ANN model; Peak Discharge Prediction; Experimental data; ANN model training and testing; FL model calibration and validation; KWA model calibration and validation; Runoff Hydrograph Simulation; ANN model training and testing; FL model calibration and validation; KWA model calibration and verification; Hydrograph Simulation at Watershed Scale; Dispersion Prediction; Experimental data; Regression-based model; Fuzzy model; References PART III - GENETIC ALGORITHMS Chapter 10 - Genetic Algorithms (GAS) Introduction; Basic Units of GA; GA Operations; Forming initial gene pool; Evaluating fitness of each chromosome; Selection; Cross-over operation; Single cut; Double cut; Multiple cut; Uniform crossing; Using sub-chromosome; Reversing; Mutation Chapter 11 - Variant of Genetic Algorithm Variant of Genetic Algorithms; Responsive perturbation algorithm; Trait-based heterogeneous populations (TbHP); Trait-based heterogeneous populations plus (TbHP+); Test Functions; Model Testing Chapter 12 - Genetic Algorithm Model Applications in Water Resources Engineering GA Application Problems; Longitudinal dispersion coefficient in natural streams; Hydrograph simulation; Watershed and hydrologic data; GA-RCM model implementation and calibration; Hydrograph predictions; Sensitivity analysis; Number of events used in calibration; Using shorter wave travel time eventsin the calibration; Using lower peak events in calibration; Hydrograph simulation using level data; Hydrograph predictions; Mean and bankfull discharge prediction; Non-linear regression method; Artificial neural networks method; Fuzzy method; Genetic algorithm; Appendix; References