Research On Confirmation Model Of Rice Producing Based On Data Fusion Technology

The geographical position of Jilin Province is unique.The Qianguoerluosi Rice has a history of more than 70 years and has been recognized as a geographical indication rice.It has become a well-known rice top grade many years ago.Due to the superior quality and reputation of geographical indication rice,the phenomenon of adulteration of geographical indication rice in the market is rampant and mixed.The current establishment system of rice confirmation is unreasonable and imperfect.There are many loopholes,Therefore,the research and establishment of a scientific and accurate rice origin confirmation technology is of great significance.This paper mainly discusses the feasibility of applying feature-level data fusion technology in the verification of origin in different regions of the province.In order to improve the accuracy of the model,a new method of multi-data fusion technology of near-infrared spectroscopy and mineral elements was proposed.The back-propagation artificial neural network algorithm was used to establish the proof model of the Qianguo origin.The research on the confirmation of the production of Qianguo rice in Jilin Province provides new ideas and methods.The main conclusions are as follows:(1)After comparing the three levels of data fusion methods,it is found that the feature level data fusion method is more suitable for the data type of this fusion,and can be combined with BP artificial neural network algorithm for modeling.(2)Qianguo rice yield confirmation model based on mineral element data and BP artificial neural network shows a high accuracy rate of 100%,and The accuracy of the test set is 92.31%,which can accurately verify the origin of Qianguo rice.(3)After five kinds of near-infrared spectroscopy pretreatment methods,the Savitzky-Golay convolution smoothing pre-processed data is used to establish BP artificial neural network model after principal component analysis(PCA).The model training set is accurate.The rate is 100% and the test set accuracy is 96.15%.(4)The BP artificial neural network model training set based on feature level data fusion technology has an accuracy rate of 100% and a test accuracy rate of 100%.Compared with the single spectral fingerprint information model(96.15%)and the mineral element model(92.31%)they increased by 3.85% and 7.69%,respectively.Feature-level data fusion technology combined with BP artificial neural network technology can accurately confirm the origin of Qianguo rice.(5)Compared with the other two single data domain BP neural network models,the model based on mineral element data and BP artificial neural network has higher accuracy and stronger generalization ability.This method proves to be feasible in the confirmation of the production of Qianguo rice.