Research On Food Safety Risk Pre-warning System Based On Big Data Mining

Predicting the food safety risk situation in a region is a hot and difficult issue for China's grass-roots food safety supervision.The unqualified rates evaluation method of the food safety in the jurisdiction area of the grass-roots regulatory authorities is easy and practicable.However,this method does not reflect the trend of some food risk factors in a region.In order to make up for this shortcoming,this Paper tries to construct the evaluation system of food risk coefficient.The evaluation system of food risk coefficient analyzes and discusses the connotation of food safety,the category of the harmful factors in food and its influence on food safety from the point of view of evaluation object and supervision,provides the basis for the construction of food risk coefficient.Based on the food risk coefficient,combined with the BP artificial neural network model,the data trend of the food risk coefficient,which is simple,objective and operable,is put forward,and the principle,index system and calculation method are put forward.The food risk coefficient is successfully applied in various fields,such as various types of food,various types of parameters of safety trend forecasting and warning,etc.,for the grass-roots food safety supervision targeted to provide scientific theoretical basis.The main conclusions are as follow:(1)The food safety risk coe fficient is built on the principle of "material determination","the data source determination",“standardize value limits " and “information comprehensive",it has the corresponding simplicity and applicability.With the deterministic risk factors in food as evaluation objects,the food safety standardize value limits as basic basis,the qualified rates and unqualified rates as data input,finally the food risk coefficient was constructed successfully.The index can reflect the risk degree of some kind of f ood or some kind of parameters to be measured well,and the ability to distinguish risk is superior to the traditional one.The index calculation is more simple,the data source is more reliable,independent of subjective evaluation thus more objective.(2)The data mining of food risk coefficient by using neural network tool in MATLAB2012 a is put forward,and the food safety risk warning system is constructed.By using a large number of food risk factors as inputs,selecting the appropriate training method,and then verifying the data with the past,updating the database in time,using the new data for the training of forecasting system,obtaining the food safety based on BP neural network data mining.The early warning method enables effective prediction o f the input batch.This method is simple and easy to operate.It is very helpful for the grass-roots food safety supervision departments to improve relevant early-warning techniques,and to support the supervisory authorities to clarify the regulatory prio rities at different time periods.Finally,the effectiveness of regulatory decision-making will be improved.