The Optimization Of Models Of Comprehensive Evaluation Of Environmental Quality With Improved Invasive Weed Optimization

Environmental quality assessment is a basic work of environmental protection to reflect the objective state of environment really. It can provide scientific basis for comprehensive management of environmental pollution, environmental planning,decision-making of environmental management organization and environmental protection policies. Therefore, the study of theory and methods of environmental quality evaluation has very important significance.To establish universal and general models, methods or formulas to any environmental system quality assessment, include a variety of water environment,water resource environment, air environment, ecological environment and so on,against the following problems in models, methods or formulas of existing environmental quality assessment:(1)Traditional index method is difficult to reflect or exaggerate excessively the role some of the indicators in the evaluation, to result its evaluation results are inconsistent with the actual situation.(2)For larger indicators,the design and computational workload of evaluation function are tremendous of some uncertainty methods, such as fuzzy evaluation method, matter-element evaluation, gray evaluation method, set pair analysis method, etc.), and there is no unified design law for the evaluation function.(3)Artificial neural network, BP neural network, projection pursuit, support vector machines and other intelligent models contain more parameters to be optimized which with the increasing of indexes increase. It leads to program complicatedly, computational workload increasedly and use inconveniently. Do this, the paper establish multiple index formulae applied to evaluation of different environment systems and forward neural network model(NV-FNN)based on the normalized indexes values that combine the thought of normalized transform with intelligent optimization algorithm on the bases of proposing multiple different indexes formulae and forward neural network model. The combination opens up new ways for the simplification, standardization and unification of models, approach and formulas of environmental quality evaluation.This paper combines the National Natural Science Foundation of China( 51179110,51209024) with technological infrastructure work of special projects( 2011IM011000), proposes various design principles and methods of reference indexes values and normalized indexes transform fomulae that applied to various environmental systems, and does normalized transform for all indexes. Then it proposes forward neural network model and 9 different universal indexes formulae applied to any environmental systems evaluation based on the normalized indexes values, which contain with W-F(Weber-Fechner)law formula, carson universal index formula, logarithmic power function index formula, power function index formula,Logistic index formula, г distribution index formula, plus a power function and type index formula, multi-parameter combinations operator index formula, quadratic function formula. This paper proposes two kinds of improved invasive weed optimization which are Invasive Weed Optimization by Immune Evolutionary Algorithm(IEA-IWO)and Invasive Weed Optimization by Immune Evolutionary Algorithm based on chaotic search( IEA-CS-IWO), for the limitations of slow convergence speed and easily falling into local minimum, and uses the improved IEA-CS-IWO for parameter optimization of universal evaluation formulae and model,and obtains 9 different indexes formulae and forward neural network model based on the normalized indexes values(NV-FNN)that can be pervasive and universal for different environmental systems evaluation. Taking has been optimized multiple indexes formulae and forward neural network model use of multiple instances evaluation of water environmental systems quality( Surface water quality,groundwater quality, marine water quality, eutrophication), water resource systems( Sustainable utilization of water resources, capacity of water resources, water security), air environment quality(air environment quality(GB3095-2012)and indoor air environmental quality(GB18883-2002)), eco-environment quality and so on, and comparing with the evaluation results of other multiple traditional evaluation methods.As a result, they have universality, versatility, feasibility and practicality has been proposed multiple formulae and forward neural network model in this paper, Thus provide a new way for comprehensive evaluation of different environmental systems quality, and can provide a theoretical foundation and reference for decisions of environmental management.The main innovated research results that have obtained in this paper are as follows:(1)Normalized indexes transform fomulae that applied to various environmental systems in this paper, and made them set up simply and easily.(2)It established forward neural network model and 9 different universal indexes formulae applied to different environmental systems evaluation based on the normalized indexes values, which contained with W-F(Weber-Fechner)law formula,carson universal index formula, logarithmic power function index formula, power function index formula, Logistic index formula, г distribution index formula, plus a power function and type index formula, multi-parameter combinations operator index formula, quadratic function formula, and made the evaluation of different environmental systems get universally, standardized, unified and simplified.(3)The universal indexes formulae and forward neural network model based on normalized transform could universally apply to qualitative indexes of any environmental systems and any number of indexes. Therefore, they had broader application range and more practical than traditional indexes formulae, BP neural network model and traditional forward neural network model and so on.(4)This paper proposed two kinds of improved invasive weed optimization which are Invasive Weed Optimization by Immune Evolutionary Algorithm(IEA-IWO)and Invasive Weed Optimization by Immune Evolutionary Algorithm based on chaotic search(IEA-CS-IWO), for the limitations of slow convergence speed and easily falling into local minimum, and tested the performance of 8 typical test functions, the result showed that the two kinds of improved invasive weed optimization not only the convergence was speeder, but also the optimization results were better than traditionalinvasive weed optimization.(5)This article first used the Invasive Weed Optimization by Immune Evolutionary Algorithm based on chaotic search for parameters optimization of environmental evaluation model and formulae.The modeling thoughts and methods combined normalized transform with optimization algorithm proposed in this paper that provided a new ideas for the analysis process of high-dimensional and nonlinear data or the establish of other models, and they would have a role of references and inspirations that simplify the modeling of multiple regression, projection pursuit regression, support vector machine regression and radial basis function neural network for evaluation and forecast of environmental systems.