The Research On The Application Of Efficient Non-linear Algorithms In Water Environmental System

Water environmental system is unordered, unstable and non-equilibrium, where usually exists the randomicity, cooperative phenomena and coherent effects. However, due to the influences of many factors, such as chronometer, weather, climate and human science, the optimization problems in water environmental system are always high-dimensional, multi-peak, discontinuous, non-convex and with noise, which is so complicated that surpasses the abilities of traditional optimization methods. In recent years, more and more scholars commit themselves to the research of optimization problems in water environment system, the theories and methods of which are of limitations in practice use and result in the enormous waste of resource and funds finally.Therefore, the paper primarily explores the application of efficient non-linear algorithms in water environmental system and puts stress on the discussion of Genetic Algorithm (GA), Simulated Annealing (SA) and Artificial Nerve Network (ANN). Focusing on the defects of the algorithms mentioning above, the improved efficient non-linear algorithms are also proposed combining Fuzzy mathematics, Simplex algorithm and wavelet analysis. Eventually, the paper solves the programming in water pollution control system, the optimization of discharge fort, multi-objective programming, identifying of water quality models parameters and the prediction of water quality using the efficient non-linear algorithms. The results show that:The solving processes of certain or uncertain water pollution system programming models are simplified greatly by using non-linear algorithms. And global optimums can be found quickly. The results don't depend on initial values if the optimization parameters are appropriate selected;The difficulties of how to select the weights of each object and avoid the errors and losses by simplifying certain object ones in traditional methods can be solved by using Fuzzy mathematic theory. Combining the advantages of non-linear algorithms and Fuzzy theory, the improved algorithm corresponds the relations between each objects and gets the global optimums of multi-objective programming in water pollution controlling system;The objective functions are frequently complex and non-linear when estimating the parameters of water quality models. Especially, when many parameters are needed estimating, there will exist many local optimums in objective function, which isdifficult to solve for traditional methods. Non-linear algorithms or improved algorithms considering the randomicity of the generation of new values and assuring the integrity of searching at some extents, will accelerate the speed of searching optimums;Efficient non-linear algorithms estimating models can raise the estimation precision and reflect the relations between historical data, which are much valuable for environmental decision department to establish the stratagem, metaphase'or long-term programming targets. Differing from traditional methods, the algorithm uses experimentation data as swatch and studies the data using the abilities of self-studying and memories, which is approximate to the swatch, distinguish the non-linear mapping between each influence factors and is a new approach to improve the precision and practicability in water quality estimation.The relationships between each factors can be corresponded, the unifying of economic benefits, environmental benefits and society benefits can be achieved and the water quality programming, evaluation and estimation will be more reliable by using efficient non-linear algorithms into water environmental system. Therefore, it's believed that non-linear algorithm would have its extensively applied prospect in the field of optimization in environmental science.