EPANET, a widely used water distribution network simulation model, is used in this study to deal with both the steady state and extended period simulation and is linked with a powerful optimization algorithm, Shuffled Complex Evolution (SCE). SCE deals with a set of population of points and searches in all direction within the feasible space based on objective function. In this present study, SCE is applied for the design of a cost effective water distribution network. The findings of this study show that SCE is computationally much faster when compared with other also widely used algorithms such as GAs, Simulated Annealing, GLOBE and Shuffled Frog Leaping Algorithms. Hence, SCE is a potential alternative optimization algorithm to solve water distribution network problems.

this paper, we examine two approaches. Runoff simulations based on an existing lumped parameter model, using mean areal precipitation as input, are compared to those based on a reformulated model, which attempts to better utilize the information available from weather radars by including the spatial variability of rainfall over the land segment area. The effect of precipitation sampling on model parameter calibration is also examined.

Markov chain Monte Carlo (MCMC) methods have found widespread use in many fields of study to estimate the average properties of complex systems, and for posterior inference in a Bayesian framework. Existing theory and experiments prove convergence of well constructed MCMC schemes to the appropriate limiting distribution under a variety of different conditions. In practice, however this convergence is often observed to be disturbingly slow. This is frequently caused by an inappropriate selection of the proposal distribution used to generate trial moves in the Markov Chain. Here we show that significant improvements to the efficiency of MCMC simulation can be made by using a self-adaptive Differential Evolution learning strategy within a population-based evolutionary framework. This scheme, entitled DiffeRential Evolution Adaptive Metropolis or DREAM, runs multiple different chains simultaneously for global exploration, and automatically tunes the scale and orientation of the proposal distribution in randomized subspaces during the search. Ergodicity of the algorithm is proved, and various examples involving nonlinearity, high-dimensionality, and multimodality show

In this paper, we take a "computational experiments" approach to robust stability analysis problems. Many robust control problems have been shown to be NP hard but in spite of this, it is important to develop effective techniques for solving them. A typical robust stability analysis problem is taken and formulated as an optimization problem to which several optimization algorithms are applied. Key words: randomized algorithms, global optimization, robust stability analysis, real parametric uncertainty 1 Introduction In the last few years, it has become increasingly clear that the problem of robust stability analysis (even for linear time-invariant finite dimensional control systems) is extremely difficult to tackle analytically. This appears to be even more so for the case of real parameter uncertainty. Results on computational complexity [5,14] of robust stability analysis provide strong support for such conclusions. The fundamental underlying reason appears to be the fact that th...

Abstract—Reliable water level forecasts are particularly important for warning against dangerous flood and inundation. The current study aims at investigating the suitability of the adaptive network based fuzzy inference system for continuous water level modeling. A hybrid learning algorithm, which combines the least square method and the back propagation algorithm, is used to identify the parameters of the network. For this study, water levels data are available for a hydrological year of 2002 with a sampling interval of 1-hour. The number of antecedent water level that should be included in the input variables is determined by two statistical methods, i.e. autocorrelation function and partial autocorrelation function between the variables. Forecasting was done for 1-hour until 12-hour ahead in order to compare the models generalization at higher horizons. The results demonstrate that the adaptive networkbased fuzzy inference system model can be applied successfully and provide high accuracy and reliability for river water level estimation. In general, the adaptive network-based fuzzy inference system provides accurate and reliable water level prediction for 1-hour ahead where the MAPE=1.15 % and correlation=0.98 was achieved. Up to 12-hour ahead prediction, the model still shows relatively good performance where the error of prediction resulted was less than 9.65%. The information gathered from the preliminary results provide a useful guidance or reference for flood early warning system design in which the magnitude and the timing of a potential extreme flood are indicated.

Abstract A reliable flood warning system depends on efficient and accurate forecasting technology. A systematic investigation of three common types of artificial neural networks (ANNs) for multi-stepahead (MSA) flood forecasting is presented. The operating mechanisms and principles of the three types of MSA neural networks are explored: multi-input multi-output (MIMO), multi-input single-output (MISO) and serial-propagated structure. The most commonly used multi-layer feed-forward networks with conjugate gradient algorithm are adopted for application. Rainfall–runoff data sets from two watersheds in Taiwan are used separately to investigate the effectiveness and stability of the neural networks for MSA flood forecasting. The results indicate consistently that, even though the MIMO is the most common architecture presented in ANNs, it is less accurate because its multi-objectives (predicted many time steps) must be optimized simultaneously. Both MISO and serial-propagated neural networks are capable of performing accurate short-term (one- or two-step-ahead) forecasting. For longterm (more than two steps) forecasts, only the serial-propagated neural network could provide satisfactory results in both watersheds. The results suggest that the serial-propagated structure can help in improving the accuracy of MSA flood forecasts. Key words neural networks; multi-step-ahead; flood forecasting; serial-propagated structure; Taiwan

The thesis will be available as a pdf-file for downloading from the institute homepage on: www.er.dtu.dk

Optimal use of the SCE-UA global optimization method for calibrating watershed models

Abstract: Hydrological models have been increasing in complexity over the years. These models rely on their calibration to simulate real world conditions as close as possible. Calibration is a tedious and time-consuming process. An auto-calibration algorithm (SCE-UA) developed by Duan et al. [1992], has been successfully used in hydrological modeling area. This is a serial algorithm and as complexity of the models to be calibrated increases the computational cost, also significantly increases. In this study, a parallel version of the algorithm developed is used for testing of two simple hydrological models. The results show that parallel version of the algorithm can be successfully used to calibrate complex hydrological models. Keywords: Shuffled-Complex Evolution Algorithm; High Performance Computing. 1.

In recent years, a strong debate has emerged in the hydrologic literature regarding what constitutes an appropriate framework for uncertainty estimation. Particularly, there is strong disagreement whether an uncertainty framework should have its roots within a proper statistical (Bayesian) context, or whether such a framework should be based on a different philosophy and implement informal measures and weak procedures to summarize parameter and predictive distributions. In this paper, we compare a formal Bayesian approach using Markov Chain Monte Carlo (MCMC) with Generalized Likelihood Uncertainty Estimation (GLUE) for assessing uncertainty in conceptual watershed modeling. Our formal Bayesian approach is implemented using the recently developed DiffeRential Evolution Adaptive Metropolis (DREAM) MCMC scheme with a likelihood function that explicitly considers model structural, input and parameter uncertainty. Our results demonstrate that DREAM and GLUE can generate very similar estimates of total streamflow uncertainty. This suggests that formal and informal Bayesian approaches have more common ground than the hydrologic literature and ongoing debate might suggest. The main advantage of formal approaches is, however, that they attempt to disentangle the effect of forcing, parameter and model structural error on total predictive uncertainty. This is key to improving hydrologic theory and to better understand and predict the flow of water through catchments.