Application Of Support Vector Machine With Genetic Algorithm In The Air Quality Assessment

With the progress of society, the economy developing rapidly, a large number of discharge of pollutants into the atmosphere which industrial enterprises generated in the production process, the air quality levels decreased, So how to make economic and environmental development harmoniously has become the focus of attention. In order to more effectively carry out the prevention and control of air pollution more effectively, it is necessary to evaluate the quality of the air scientifically and reasonably, to determine their respective levels and identify the important factors, as a result,it can put forword reasonable suggestions and planning.Air quality assessment models influenced by some subjective factors,it cannot well simulate the uncertain relationship between the pollutant concentration and output level of air quality, making predicted and actual values greater degree of difference. Support vector machine is a new learning machine and it is developing rapidly in the20th century, it is based on statistical learning theory and structural risk minimization principle. For processing the nonlinear relationship between variables, through a nonlinear kernel function support vector machine will make the variable from the low dimensional feature space was mapped into a high dimension space, thus the problem into solving linear equations, the problem is transformed into solving linear equations, a good estimate of the independent variables and the dependencies between the dependent variable. The SVM-based method has been compared with other traditional statistical methods and has shown good result and generalization ability, In financial services, network detection, text and handwriting recognition, bioinformatics and other fields has been successfully applied. In this paper, the support vector machine method is applied to the air quality rating, achieved good prediction effect. For the parameters of support vector machine model is given by the user in advance and there is the lack of theoretical guidance. This paper using genetic algorithms for parameter optimization improves support vector machine, it is easy to operate,without making any assumptions, Only need to select a potential solution set initial population, genetic algorithm can be self-organizing, adaptive manner heuristic parameter optimization, which has strong robustness. In this paper, the genetic algorithm dynamically determine the crossover probability and mutation probability, need only give the crossover probability and mutation probability of the initial value and the lower limit, it can automatically adjust the probability with evolution algebra size, and it has good adaptability and flexibility. Compared with the traditional support vector machine,BP neural network and RBF neural network, Support vector machine based on genetic algorithm shows better classification results and has the advantages of good stability and generalization ability.