The Research On Evaluation And Optimization Of Forest Road Network

Sustainable development is the necessary develop mode of the human society, forestry sustainable development is a specific application and reflective of the sustainable development theory. The traditional forest road network optimization theory, the evaluation index system and the method can’t adapt to the need of sustainable development of forest region, therefore, it has important significance and application value for the optimization and the evaluation research on forest road network to construct a complete evaluation index system of forest road network which can adapt to the need of sustainable development. This dissertation put support vector machine and ant colony algorithm into the evaluation on the forest road network and optimization baesd on the evaluation index system constructed in the third chapter. Research results show that these two methods can completely be used in the evaluation and optimization problems on the forest road network, and the performance is better than the methods commonly used, the result is reasonable and feasible.The main research contents of this dissertation are:(1) Referenced the construction method and research results of evaluation index system of the forest road network at home and abroad, the evaluation index of sustainable development of forest road network were screened by frequency analysis, delphi method and analytic hierarchy process, each of the index weight was calculated, and finally the complete evaluation index system of forest road network was constructed based on tne sustainable development theory.(2) On the basis of comparison and analysis the quality of several common support vector machine (SVM) classification algorithm and kernel function, a kind of forest road network evaluation model was established based on support vector machine, used the One-Against-The-Rest classification algorithm and radial basis kernel function based on the evaluation index system of forest road network. The libsvm package was used in the kernel function parameter selection, sample training and testing. The public welfare forest road network in Guangzhou and commercial forest road network from YueYang forest and paper Co., LTD were used for empirical research respectively, compared to the BP neural network, fuzzy comprehensive evaluation and k nearest neighbor classification algorithm, the results show that...