multi objective optimization

Urmia Lake has faced a water shortage crisis and a sharp decline in water level due to decreasing precipitation and increasing water consumption in the agricultural sector. This study aims to develop a hydrological model to simulate the daily inflow of Bukan reservoir and simulate the water balance components for the upstream sub-basins of this reservoir through a new approach. In this study, two separate storages are considered for daily simulation of snow and root zone soil moisture, and multi-objective calibration (MOPSO) is applied to simultaneously maximize the simulation accuracy of observed river discharge and logarithm of river discharge. The results showed a relatively little trade-off between maximizing the two objective functions to estimate the optimal value of the calibration parameters. In other words, there is no significant error in the model that simultaneously prevents improving both objective functions. Results showed that the average Nash-Sutcliffe criterion for simulation of river discharge and the logarithm of river discharge were respectively 0.43 and 0.63 in the calibration and 0.54 and 0.57 in the validation step. Water balance modeling indicated that precipitation and inflow to the reservoir were decreased respectively by 32% and 40% in severe drought years and soil moisture and snow accumulation were reduced by 32% and 158%, respectively, compared to the long-term period. The developed model can predict the effects of climate change and climate variability on Bukan reservoir inflow and optimally allocate reservoir water to satisfy water demand.

Water distribution network one of the most important and most sensitive urban infrastructures which can be recently with regard to population growth and consumers '' needs to increase with challenges such as reducing the pressure and discharge that are all due to unsuitable design and based on economic goals. In order to overcome these problems when designing an urban water distribution system should always be considered reliability of the network. In this research, to optimize the water distribution networks, two main objectives of cost minimization and maximization of network reliability were considered. To calculate network reliability, the Tondini''s Resilience Index and its hydraulic simulation EPANET 2.0 model were used. Then using the second edition of the evolutionary algorithm based on the strength of the Pareto (SPEA2) and creating a dynamic connection with the EPANET 2.0 hydraulic model in the MATLAB software environment, optimizing the multi-objective of the four water distribution networks including Two-Loop, Kadu, Hanoi networks and D zone of Mashhad. Simultaneous optimization of two main objectives including cost minimization and maximization of the Tondini''s Resilience Index led to the production of optimal solutions in the form of the Pareto front. An optimal solution, called Point C, was introduced using Young''s bargaining method from the final Pareto front of in each of the networks. Selected C solution in two-loop,Hanoi and Kadu networks increased 22.91, 17.13 and 7.41 precent, of the network average pressure compared to its lowest cost in this study (point A). Also, the selected C solution in D-zone network of Mashhad, with an increase of 4.23 precent of the network average pressure compared to the initial design of the consulting company (point D), illustrate that the solution designed by the consulting company would be a dominated solution under the final Pareto front of this study. In this research, the Tondini''s Resilience Index illustrate that, based on increasing nodal pressure, it has the ability to increase the reliability of the network, which This causes the network to be in critical condition or failure of the pipes, with high reliability, providing adequate pressure and discharge in other nodes. Also, the satisfactory performance of the SPEA2 multi-objective algorithm in providing the optimal Pareto front for the issues indicated showed that the design pattern developed in this study could be to provide an optimal set of solutions to the employer to select the points in which each two factors of cost and reliability are combined in a favorable situation.

Multi-objective problems rarely ever provide a single optimal solution, rather they yield an optimal set of outputs (Pareto fronts). Solving these problems was previously accomplished by using some simplifier methods such as the weighting coefficient method used for converting a multi-objective problem to a single objective function. However, such robust tools as multi-objective meta-heuristic algorithms have been recently developed for solving these problems. The hedging model is one of the classic problems for reservoir operation that is generally employed for mitigating drought impacts in water resources management. According to this method, although it is possible to supply the total planned demands, only portions of the demands are met to save water by allowing small deficits in the current conditions in order to avoid or reduce severe deficits in future. The approach heavily depends on economic and social considerations. In the present study, the meta-heuristic algorithms of NSGA-II, MOPSO, SPEA-II, and AMALGAM are used toward the multi-objective optimization of the hedging model. For this purpose, the rationing factors involved in Taleghan dam operation are optimized over a 35-year statistical period of inflow. There are two objective functions: a) minimizing the modified shortage index, and b) maximizing the reliability index (i.e., two opposite objectives). The results show that the above algorithms are applicable to a wide range of optimal solutions. Among the algorithms, AMALGAM is found to produce a better Pareto front for the values of the objective function, indicating its more satisfactory performance.

Optimization of reservoirs operation is one of the most important tasks in the field of water resources management. In fact, vital requirement for beneficial use of water and energy resources clear the necessity of doing integrated planning and right operation of dams. recently, research has been made focusing on a shift from traditional single objective models to multi–objective models for the planning of multiple reservoir systems in a river basin. In this study the three objectives of meeting irrigation and environmental demand, flood control and recreation (sometimes in conflict with each other) are referred to for a two reservoir system by Goal Programming. Within this framework, the mathematical model of two reservoirs system in Sefidrud watershed (Northern Iran) with the three objectives is formulated and the system parameters and decision variables are defined. The problem involves finding desired water releases from each reservoir in the system in order to satisfy the multiple objectives. With comparing results of optimization models of this study, the model with the higher reliability indices was chosen as the best model. Due to the considerable advantages of linguistic rules in better inferring and interpreting the systems, an adaptive neural based fuzzy inference system (ANFIS) approach is used to consider uncertainties and to achieve a general method for multipurpose multi reservoir systems. The results of the Adaptive Neural Fuzzy Inference System (ANFIS) models shows that they can be applied successfully to provide high accuracy for the management of the reservoir systems.

In water distribution networks the initial and rehabilitation design are usually perform separately. However, it seems that the influence of the initial design upon the future condition of the network performance during operational years and rehabilitation activities are undeniable. Therefore, by combining the initial and rehabilitation designs, a new method is presented in this paper. This method called Dynamic Design of water distribution networks is capable of introducing cheaper and more reliable long term designs in comparison with normal initial design and rehabilitation design of networks. To assess this method, a fuzzy reliability index is introduced. Then by developing the multi objective version of the honey-bee mating optimization algorithm and applying it on two sample networks, final results of the multi objective dynamic design method is presented. Finally, this paper showed the positive performance and influence of dynamic design method on decreasing the design costs and increasing system reliability.