基于BP神经网络的鸡精货架期模型研究  

Research on Model for Chicken Essence's Shelf Life Based on BP Neural Network

作　　者：肖立中 费玉刚 田怀香 张雅敬 李文举 王广仲 

机构地区：[1]上海应用技术大学计算机科学与信息工程学院,上海201418 [2]上海应用技术大学香料香精技术与工程学院,上海201418

出　　处：《中国调味品》2017年第7期16-19,24,共5页China Condiment

基　　金：上海应用技术学院协同创新基金--跨学科;多领域合作研究专项(XTCX2015-13);2015年度"创新行动计划"地方院校能力建设项目(15590503500)

摘　　要：鸡精货架期是鸡精质量评价的重要指标,对鸡精存储和品质控制具有重要意义。为了快速、准确地识别鸡精的货架期,提出了基于BP神经网络的鸡精货架期模型。结合GC-MS和人工感官评价获取了鸡精的货架期数据,针对众多的GC-MS数据,通过计算拟合决定系数和归一化对数据进行了特征选择和预处理。通过BP神经网络结构、参数和网络函数的优化,得到了高质量的模型。经过与化学动力学模型的比较实验表明提出的模型准确有效。The shelf life of chicken essence is an important index for quality evaluation and it is also be regarded as a significant factor for storage and quality control. In this paper, a model for shelf life of chicken essence based on BP neural network has been put forward for a quick and easy recognition for the shelf life. Obtain the shelf life data of chicken essence through the combination of GC-MS and manual sensory evaluation. As for the numerous GC-MS data, feature selections and pretreatments of the data are conducted by fitting coefficient of determination and normalization. High-quality model is obtained through BP neural network structure, parameters and the optimization of network function. The model presented in the paper is indicated to be accurate and effective upon comparative experiment with a chemical kinetics model.

关 键 词：鸡精 货架期 BP神经网络 化学动力学模型 

分 类 号：TS264.3[轻工技术与工程—发酵工程]