The Study On Detecting Method Of Corn Aging

Corn will be aging with the extension of time during the storage process, which caused decline in quality. It is not appropriate for rations, so accurate judgment of corn aging is of great significance as to control the grain reserving and guarantee food security. At present, the national standard method of coin aging estimation is to make a titration experiment of free fatty acid extracted by ethanol, or to obtain its degustation grade value by cooking. In this paper, the purpose of the study is to find a more simple and sensitive method to solve the above two methods'bigger subjective error.In this study, this paper found a mixed solution with the suitable ratio between Aerosol OT and isooctane to extract free fatty acids, instead of the ethanol in traditional method, and used a color sensor to reading color values, instead of the traditional method of visual measurement. The trial confirmed the testing conditions for extracting corn addition is of 0.4 g/10ml, extraction time of 15 min, and centrifugal time of 1 min. This paper also tries to use near-infrared analysis to measure the corn taste score, so as to replace the traditional methods of sensory evaluation tests. Test samples content is determined by 12g.Through studying the relaticity of color and free fatty acid value, near-infrared analyzer's original value and taste value score, we can make the following conclusions:1. The blue color sensor values and free fatty acid value of the relatively good correlation, regression equation obtained R2 = 0.9798, the regression equation to predict corn aging measurement accuracy of the situation reached 83.3%. 2. The establishment of BP neural network processes the values of B, G, Y, R color obtained by the color sensor, the network forecast aging of corn an accurate rate of 96.7%.The network nodes for input layer 4, hidden nodes to 30, the output layer for a number of nodes, input layer to the hidden layer, the transfer function of the number of S-type selection (sigmoid) function, the hidden layer to the output layer selection of the linear transfer function of the transfer function, the training of 80 samples of the initial study was 0.8, momentum factor of 0.9, network error is less than 0.001.3. By using artificial aging from the same species of maize samples, we established the multiple linear relationship of near-infrared analysis of the original test score values and taste value to predict corn aging.This method can be applied to same variety of the corn samples testing of the accuracy of 90% and different varieties of the corn samples testing of the accuracy of 77%.4. Through the compare among different kind of methods, BP neural work is selected as to detect the corn aging extent finally. Based on the Matlab tool, design and develop the corn aging detecting system. The system composes from three modules, which are neural work training module, color eigenvalue collecting module and aging extent estimating module. The system realize the corn aging estimation export automatically.