Research On Intelligent Cognition Model And Operation Mechanism Of Greengage Grade Based On Deep Learning

Chinese fruit planting area and yield are in the forefront of the world,but the level of commercialized processing after fruit picking has become one of the main factors that restrict the added value of domestic fruits and the competitiveness of the international market.The realization of automatic grading of fruit grade has become a necessary prerequisite for the modernization of the domestic fruit industry.As a medicine and food resource with multiple health functions,greengage is deeply loved by the masses.Aiming at the drawbacks of the traditional fruit grading open loop cognitive system,once the characteristic space and classification criteria are established,they will no longer update and the insufficiency of the convolution neural network architecture,the feature space redundancy and the poor generalization ability of the classifier as well as the problem of the posterior probability evaluation method cannot evaluate the cognitive results online.Based on the deep learning,this paper explores an intelligent feedback cognitive method for greengage grade with the constraints of the entropy measure of cognitive results,in order to improve the recognition rate of the greengage grade.The main work of this paper is list as follows:(1)In view of the lack of greengage grade information modeling theory,based on the rough set theory,the decision attribute information of the greengage grade image is introduced,under the uncertainty of the finite domain,from the point of view of information theory,the intelligent decision information system model of greengage grade cognitive intelligence is set up with the unstructured and multi-layer dynamic characteristics characterized by information completeness evaluation index.(2)In view of the description of the feature space and the cognitive criteria of the greengage grade image is not unified,the characteristic spatial data structure and classification criterion of the green plum grade image from the whole to the local feature characterization mapping system are established based on the Architecture Adaptive convolution neural network and the integrated random weight vector function connection network classifier.(3)In view of the problem of the posterior probability evaluation method cannot evaluate the greengage grade cognitive results online,a method of evaluating the entropy function of the green plum grade image is defined based on the generalized error theory and the generalized entropy theory,which provides a quantifiable basis for the intelligent feedback cognitive operation mechanism.(4)In view of the problem that the intelligent feedback cognitive operation mechanism of the uncertain cognitive result entropy measure index has not achieved consensus,a dynamic feedback cognitive intelligent operation mechanism based on the uncertainty of the performance measurement index of the greengage level cognitive results is established,and the recognition feature efficiency and the level of interlayer cognitive feature level are recognized in the self-optimizing adjustment layer to update the unstructured multi-level simple feature space and the classification cognition criterion of the next cognitive process.In order to validate the superiority of the proposed greengage grade intelligent feedback cognitive model,1008 images of greengage are selected as sample library,and the feasibility and effectiveness of this method are verified by MATLAB simulation.The experimental results show that the average cognitive accuracy of the 500 random sampling experiments is 98.15%,and the performance is better than the other cognitive methods,which lays the foundation for the "intelligent machine classification" of the green plum grade...