Soil Fertility Evaluation Research Based On PCA And SVM

Soil fertility is a comprehensive evaluation of soil factors on certain geographical areas, it is also called the soil fertility evaluation. At present, there are many soil fertility evaluation methods, mainly include the analytic hierarchy process, judging index method, grey correlation analysis method and fuzzy comprehensive evaluation method, etc. In recent years, the development of computer and information technology makes the neural network and SVM become a hot spot of the soil fertility evaluation.This article first introduces the level of soil fertility evaluation methods commonly used at home and abroad, and analyze the deficiency of the existing methods. According to the characteristics of the data of soil, propose an evaluation method based on the PCA and SVM. Then briefly discusses something important about SVM theory. Using membership function to calculate the weight of each evaluation index, overcome the traditional method of subjective set weight. Calculate the soil fertility level as the label of sample. Building the appraisal model of PCA-SVM, and applied to soil grade evaluation experiment. Compared with single SVM model, BP neural network evaluation model, apriori evaluation model to analyze the feasibility of the model according to the result of the experiment.Experiments show that the PCA-SVM evaluation model can not only subtract the related data of soil fertility and reduce data redundancy, reduce the complexity of the data, but also do better in terms of classification accuracy and stability, reduce the effect of subjective factor, provide a new choose of thought for soil grade evaluation.