基于神经网络的大豆油贮藏过程中品质预测分析  

Predictive Analysis of Soybean Oil Quality in Lab Storage Based on Neural Network

作　　者：曹冬梅[1] 张东杰[1] 

机构地区：[1]黑龙江八一农垦大学食品学院,大庆163319

出　　处：《中国粮油学报》2011年第6期106-109,共4页Journal of the Chinese Cereals and Oils Association

基　　金：黑龙江省教育厅科学技术研究项目(11521197);黑龙江省自然科学基金重点项目(ZJN0505-01);黑龙江省科技公关重点项目(GB05C10101)

摘　　要：根据GB 1535—2003中华人民共和国大豆油国家标准,在室温、避光储存270天的条件下,对同一批次的5个样本大豆油每15天检测其过氧化值(POV)和酸值(AV)作为训练样本,基于反向传播算法的人工神经网络建立了大豆油贮藏品质的模型,预测大豆油的过氧化值和酸值在贮藏过程中随时间变化的规律。通过过氧化值、酸值的预测值与实测值的拟合曲线和误差曲线比较及试验验证,确定了预测模型具有较高的评价精度、较低的误差率。According to GB 1535-2003 National Standard of the People’s Republic of China for Soya Bean Oil,5 samples in the same batch of soybean oil stored for 270 days and their POV and AV were used as the training sample in room-temperature and light-free conditions,and the values of room temperature storage soy oil was detected in interval 15 days.The model of soy oil storage based on the back-propagation algorithm of artificial neural network was established to forecast the POV and AV change with time during storage.The model was verified by the experimentation and the contrast of matched curve and deviation curve of predicted value and measured value,and proved to have a higher accuracy and a lower error.

关 键 词：神经网络 BP算法 预测分析 过氧化值 酸值 

分 类 号：TS227[轻工技术与工程—粮食、油脂及植物蛋白工程]