Support Vector Machine And Its Application To Regional Water Resources Sustainable Utilization

Water resources sustainable utilization is not only the foundation but also the core of regional economic and social sustainable development. As the two important contents, the assessment of water resources sustainable utilization and carrying capacity prediction's purposes are to accurately reflect the status of water resources utilization, predict future's trend. Currently, the main problem is that there has not a comparable model recognized by general. Choosing mathematical models has great arbitrary when the departments carry out the assessment and prediction. Although the traditional methods dominate in practical application, they still have many deficiencies.Support vector machine is a new algorithm rising in recent years, and it becomes the scientific research cutting-edge of the complex non-linear and artificial intelligence. It is gradually carried out in a wide range of many research fields and studies because of its outstanding performance between classification and regression. In this paper, the predecessors'work was summarized. Its application to water resources sustainable utilization was lucubrated, especially to the assessment of water resources sustainable utilization and carrying capacity prediction. A parameter optimizing method based on shuffled frog leaping algorithm was proposed to improve the model's accuracy. The main contents of the paper are as follows:(1) The current methods of the assessment of water resources sustainable utilization and carrying capacity prediction were briefly summarized. There existed some inadequacies in the current methods, so the algorithm of support vector machine was introduced. Firstly, a simple review of support vector machine'research status and development process was made. Secondly, its advantages and shortcomings were summarized.(2) The theoretical basis of support vector machine was outlined: machine learning and statistical learning theory. Not only its optimal classification surface but also its classification and regression algorithms were discussed in detail. It's conducive to understand deep support vector machine. (3) The types and parameters of kernel function were important factors of affecting support vector machine's performance. Several common optimizing methods of kernel parameters were described, then a parameter optimizing method based on shuffled frog leaping algorithm was proposed. Its superiority was proved by some tests.(4) The assessment model for water resources sustainable utilization and the prediction model for carrying capacity were established by the classification and regression algorithms of support vector machine. Because there were lots of factors, so rough set theory was used to reduce the properties, extract main factors, remove irrelevant information, and reduce the workload. In addition, the best combination of parameters was selected by the parameter optimizing method based on shuffled frog leaping algorithm. The accuracy of the model was improved.(5) Water resources were the major constraint factor for the development of Minqin oasis. After the datas and information were collected, two models were used to the practical application of Minqin oasis. Many important informations such as the current level of water resources sustainable utilization and the situation of water resources carrying capacity were obtained, which provided the basis for the future social's and economic's development of Minqin oasis.