Research And Application Of Vegetables Safety Early Warning System Based On Weighted Support Vector Machine

Vegetables safety plays an important role in the development of the society. On vegetables safety risk prevention, the establishment of an effective vegetables safety assessment model is an important fortress. In this paper, a method of vegetables safety assessment and prediction based on weighted support vector machine is proposed, and then set up vegetables safety early warning system.According to the characteristics of vegetables safety early warning, in this paper through study and comparison analysis of various evaluation prediction techniques, such as hierarchical analysis method, neural network, using support vector machine (SVM) as vegetables safety assessment and prediction model of the core algorithm, to solve the vegetables safety evaluation in small sample, nonlinear and high dimension problems. At the same time, weighted support vector machine is used to form a vegetables safety evaluation model, which can be used to reduce the prediction tendency caused by the difference of the number of different types of samples. According to the Likert five point scale and practical investigation, the vegetables safety status is divided into five categories, and the support vector machine is the two kind of classification model. Therefore, this paper introduces the directed acyclic graph (DAG SVMs) multi class classification method, vegetables safety assessment and prediction of the five class classification model is constructed.This paper first select "very safe" and "dangerous" two kinds of samples, and selected BHC, phorate, of parathion and other eight safety indicators constitute spinach safety assessment and prediction model. Then the samples were be pre-processing, namely data format conversion and data normalization, as the training samples of support vector machine. After a trial analysis of several common kernel functions, this paper chooses Gauss radial basis kernel function as the kernel function of the vegetables safety evaluation model, and by using cross validation and grid search method, the parameters of the penalty factor c and Gauss radial basis kernel function parameter g were optimized in a certain range, and the optimal parameters were obtained (c, g), and through "very safe" and "dangerous" two kinds of training sample proportion, respectively for the weighted calculation, to., build a "very safe" and "dangerous" binary class classification vegetables safety assessment and prediction model. Then, the other two categories, such as "very safe" and "dangerous", can estimate the model according to the above steps and one one. Finally, the paper uses the directed acyclic graph (DAG-SVMs) multi class classification method, on the basis of the assessment prediction model of the two types of vegetables safety, to construct a vegetables safety assessment model to support the early warning of vegetables safety.Vegetables safety early warning system developed in this paper has been tested by the vegetables and Drug Administration of a prefecture level city, the effectiveness of the vegetables safety assessment model based on the weighted support vector machine is verified, which provides an effective basis for the early warning of vegetables safety.