CAJ | 학술논문

速冻是鲜食玉米加工的关键环节，速冻时间的合理控制是决定鲜食玉米质量的关键因素。基于BP神经网络，建立了以玉米尺寸、形状系数、密度、比热容、导热系数、环境温度、初始温度、冻结温度为输入量，玉米速冻时间为输出量的鲜食玉米速冻时间预测模型，并与传统数值计算模型进行比较。结果表明：基于BP神经网络训练得到的预测值与实际值间的最大误差（约为1?23％）要略小于传统数值模型模拟所得值与实际值间的最大误差（约为3?66％）。表明基于BP神经网络的模型对鲜食玉米速冻时间的预测更加准确。
속동시선식옥미가공적관건배절，속동시간적합리공제시결정선식옥미질량적관건인소。기우BP신경망락，건립료이옥미척촌、형상계수、밀도、비열용、도열계수、배경온도、초시온도、동결온도위수입량，옥미속동시간위수출량적선식옥미속동시간예측모형，병여전통수치계산모형진행비교。결과표명：기우BP신경망락훈련득도적예측치여실제치간적최대오차（약위1?23％）요략소우전통수치모형모의소득치여실제치간적최대오차（약위3?66％）。표명기우BP신경망락적모형대선식옥미속동시간적예측경가준학。
Freezing is a key link of fresh corn processing. The reasonable control of freezing time is the key factor to determine the quality of fresh edible maize. In this study, we established a fresh corn freezing time prediction model which took corn size, shape coefficient, density, specific heat, thermal conductivity, environmental temperature, initial temperature and freezing temperature as in?put and corn freezing time as output based on BP neural network, and compared this model with the traditional numerical literature model. By comparing the value of the two models, we found that the maximum error between BP model calculations and experimental measurements ( about 1?23%) was slightly less than the maximum error between numerical model calculations and experimental meas?urements ( about 3?66%) , which showed that greater accuracy was provided by the fresh corn freez?ing time prediction model based on BP neural network model.