摘要
采用L4(23)正交法,对雪菜腌制过程中影响亚硝酸盐含量变化的盐度、发酵方式和品种等因素进行研究,并建立了2个数学模型.结果表明:利用"最小二乘法"进行估计的线性效用估计模型,估计结果同控制亚硝酸盐产生的实际情况相吻合.利用"Matlab"软件建立神经网络模型,采用"Levenberg-Marquardt"训练方法,进行180次训练,就达到10-7精度.说明控制亚硝酸盐产生神经网络模型较线性模型更具有优越性.
The effect of some factors including fermentation method, the content of salt and the species on the content of nitrite during the process of pickling of potherb mustard was investigated by using the L_4(2~3) orthogonal experiment. Two mathematical models were provided. The results showed that one is a valid linear model that was estimated according to least twain-multiplication. The findings attained by this model was consistent with what were found in the process of pickling of potherb mustard. The other is a neural network model, which, combined with 'Matlab' software and 'Levenberg-Marquardt' exercitation means, precision arrived at 10^(-7) through 180 exercitations. It showed that the latter was superior to the former in the process of pickling of potherb mustard.
出处
《浙江大学学报（农业与生命科学版）》
CAS
CSCD
北大核心
2004年第5期515-518,共4页
Journal of Zhejiang University:Agriculture and Life Sciences
基金
浙江省自然科学基金资助项目(302418)
国家教育部科学技术研究重点资助项目(03052).
