System Research Of Artificial Forest Productivity Evaluation Based On Support Vector Machine

In recent years, with the rapid development of China’s economic and social, informatization level continuously improve in all walks of life, forestry information construction has also made considerable progress. Fertility data is one of the most important data type in the forestry, analysis of such data and evaluate the soil fertility level has become an important part of daily work in the forestry sector.Support vector machine (SVM) has become a hot topic of research as a new machine learning method, which has the advantages of small sample learning, fast convergence speed, good generalization ability and be able to get a global optimal solution, and widely applied to the field of pattern recognition, classification analysis. Rough set (RS) is a mathematical tool that deal with uncertainty and incompleteness. It has certain advantages in dealing with large amounts of data, eliminate redundant information and dealing with uncertainty information. Based on study on the algorithm of support vector machine, combines rough set theory, through the experiment in the evaluation of plantation soil fertility level, this paper verified feasibility of support vector machine (SVM) model, and finally apply it to the actual project of plantation productivity grade evaluation system.This paper first introduces the level of soil fertility evaluation methods commonly used at home and abroad, and analyzes the deficiency of existing method, combining with the characteristics of soil fertility data, puts forward the classification method based on rough set and support vector machine (SVM);Then briefly discusses the theoretical basis of support vector machines (SVM), in-depth study of the main method of support vector machine (SVM) that solve practical problems, given plantation soil fertility level of the RS-SVM evaluation model; Then RS-SVM evaluation model is applied to the plantation productivity grade evaluation experiment, and compared with single SVM evaluation model and BP artificial neural network method, according to the result of application analysis of the feasibility of the model; Finally on the basis of the research, design and development plantation soil fertility evaluation information system based on. NET platform.Experiments show that compared with single SVM evaluation methods, RS-SVM model reduce the space and time complexity of the algorithm and ensure evaluation precision at the same time, so it improve the training efficiency. Meanwhile it has a higher evaluation accuracy than the BP artificial neural network. In the Actual work of soil fertility evaluation, it is very difficult to take sample and measure for each area, in the large-scale soil fertility evaluation, especially under limited samples, using RS-SVM evaluation method evaluate soil fertility is a feasible solution.