| Water is the source of the lives, and the material basis of the human beings' survival and the social and economic development. Water resources can guarantee the sustainable development. The sustainable water resources development is involved in population resources, environment and society. With the rapid population increase, the social and economic development, and the people's living standard improvement, the quantity and quality of water requirement is becoming more and more higher, but the available freshwater is limited. Additionally the pollution and waste of water resources has sharpened contradiction between water supply and water requirement, which becomes one of the key obstacles of social development. The optimal allocation of water resources is an effective way of solving the problem. So the study on optimal allocation of regional water resources is important and meaningful to the sustainable development of society, economy and environment.In this dissertation, the research present situation about water resources optimization is concluded and discussed. The optimal allocation model of water resources facing sustainable development is built up. Based on the study of modified Genetic Algorithms, the solving method and ideas of the allocation model is put forward. Further more, the application of Artificial Neural Networks and Grey system in the forecast of water requirement is studied. In the end, the water resources optimization model is certified taking JiYuan city for an example. The concrete contents are as follows:(1) The basic theory and realizing techniques of Genetic Algorithms are systematically analyzed, and the shortcomings of GA are deeply analyzed, then some new modifying methods are presented. According to the function of different genetic operators during the evolution process, and the ideas of more excellent individual with less probability of crossover and mutation, the dynamic probabilities of crossover and mutation are given. Taking Simulated Annealing as a genetic operator realized the combination of the local searching ability of SA and global searching ability of GA. A new Genetic Simulated Annealing Hybrid Algorithms are designed and tested by a nonlinear functionoptimization. The different combinations and their effect of genetic operators of rank-based selection, arithmetic crossover, uniform mutation, non-uniform mutation, self-adapting mutation and et al are discussed. The result indicates that the hybrid algorithms based on rank-based selection, arithmetic crossover and uniform mutation is more effective.(2) The basic optimization theory of water resources is systematically discussed. Based on sustainable development theory, the recognition and measurement of social, economic, and environmental benefit are studied, and a concrete expression of the three objects is given. Considered the different features of water users and water sources, the concepts and calculation methods of water users uniform coefficient and water sources supplying coefficient are presented, which is embodied as a restriction condition. The large-scale multi-objective optimization model of regional water resources is set up, with a maximum objective of integrated benefit of society, economy and environment. On the basis of analyzing characteristics of large-scale, multi-objective, nonlinear of the allocation model and their concerning solution, two kinds of methods based on GA to resolve the model are discussed, which can enhance the theory research of large-scale multi-objective optimization and expand the new applicant fields of GA.(3) Forecasting methods and influential factors of water requirements are concluded and analyzed. The models of Artificial Neural Networks (ANN), Grey Forecasting (GF) and Exponent Smoothing (ES) are deeply discussed. With more attention paid on ANN Bp model, some modifying methods are presented and used, such as learning pattern of batching, dynamic adjustment of studying rate and inertial momentum, and auto-selection of parameters. The...
|
PDF Full Text Request