陈 华^1，2，孙恢礼^1﹡，易湘茜^1，2，3，陈 忻^1，2，4
（1 中国科学院南海海洋研究所，中科院海洋生物资源可持续利用重点实验室，广州 510301；
2 中国科学院研究生院，北京 100049；3 广西中医学院药学院，南宁 530001；
4 佛山科学技术学院理学院，佛山 528000）

摘要：介绍了人工神经网络在海洋动物蛋白酶解、分离提取、灭菌包装、成分分析、浓度测定、感官评价等环节中的应用情况，提出应结合模糊数学、数学逻辑与拓扑数学等方法探索更优的模型，以最大限度地克服其固有的缺点和不足，并应用到原料筛选、加工过程监控、故障检测、多组分分析等各个环节之中。

关键词：人工神经网络；海洋动物蛋白；水产品加工；模拟；预测；优化

Application of artificial neural network models in marine animal protein processing
Hua Chen^{1, 2}, Huili Sun¹, Xiangxi Yi ^{1, 2, 3}, Xin Chen ^{1, 2, 4}
( 1 South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou 510301, China;
2 Graduate School of the Chinese Academy of Sciences, Beijing 10049, China;
3 Guangxi University of Traditional Chinese Medicine, Nanning 530001, China;
4 Department of Chemistry and Chemical Engineering, Foshan University, Foshan 528000, China )

Abstract: Artificial Neural Network (ANN) has been used as powerful tools in non-linear system, particularly in technics optimization, process simulation, risk evaluation and prediction in various fields. However, the application of ANN in food science was still in primary stage of development. Marine animal protein processing involved in hydrolysis, quality control, component analysis and evaluation was often a non-linear and non-steady-state system. And the application of ANN can reduce workload, shorten development cycle time and save cost. So in this article, the application of ANN in marine animal protein processing was described briefly. It focused on modeling, predicting and optimizing of enzymatic hydrolysis, separating bioactive components, sterilizing, packing, analyzing ingredients, measuring metal concentration, evaluating the flavor or taste of products and so forth. After considering the disadvantages of ANN, it was put forward that ANN should be combined with fuzzy mathematics, mathematical logic, topology and other calculating methods and used in more aspects, such as filtering raw materials, controlling process, detecting malfunction and analyzing ingredients synchronously in future.

Key words: Artificial Neural Network (ANN); marine animal protein; aquatic product processing; model; predict; optimize

参考文献

[1] 周开利,康耀红.神经网络模型及其MAT LAB仿真程序设计[M ].北京:清华大学出版社,2005.
[2] 李琳,赵谋明,张黎.利用人工神经网络优化制备鳙鱼抗氧化肽[J].四川大学学报:工程科学版,2006,38(1):80-85.
[3] 叶晓,黄小葳,俞军,等.基于BP网络建立姬松茸多糖超滤分离模型[J].化学研究与应用,2006,18(9):1120-1123.
[4] 周大鹏,李谦,卢凤莉.超临界流体萃取过程混合模型的建立[J].河南大学学报:自然科学版,2006, 6(3):42-46.
[5] TORRECILLA J S, OTERO L, SANZ P D. Optimization of an artificial neural network for thermal/pressure food processing: Evaluation of training lgorithms[J]. Computers and Electronics in Agriculture, 2007, 56(2): 101-110.
[6] 徐胜,张金泽,薛毅.人工神经网络在无菌包装系统包装材料杀菌效果评估的应用[J].食品与发酵工业,2006,32(3):55-59.
[7] 夏远景,陈淑花,薛路舟,等.超高压处理牡蛎灭菌实验研究及人工神经网络模拟[J].现代食品科技,2009,25(5):530-533.
[8] 唐明翔,陈海元,杨公明,等.食品添加剂高效液相色谱分析的神经网络建模研究[J].食品科学,2007,28(5):77-80.
[9] 王志有,于洪梅,李井会,等.BP人工神经网络紫外分光光度法同时测定三种氨基酸[J].生物数学学报,2005,20(2):240-244.
[10] 于洪梅,王志有,李井会,等.偏最小二乘法BP网络光度法同时测定复方氨基酸中色氨酸、酪氨酸和苯丙氨酸[J].检验化验-化学分册,2008,44(5):411-413.
[11] 殷勇,易军鹏,李欣,等.食品中铜铅镉锌同时测定的神经网络方法研究[J].食品科学,2005,26(8):271-274.
[12]殷勇, 陈朝魁, 易军鹏. 基于小波包神经网络的食品中锌、铁、锰元素电化学同时检测方法研究[J].食品科学,2008,29(6):342-345.
[13] 王奉堂.用人工神经网络预测食品感官性的新方法[J].冷饮与速冻食品工业,1996(1):19-25.

基金项目：国家“863”高技术研发项目（2007AA091602）；国家科技支撑计划项目（2008BAD94B08）；中科院重要方向项目（KZCX2YW209）；粤港招标项目（2007498611）
作者简介：陈华（1982—），女，博士生，研究方向：海洋生物高值化利用。E-mail: alicepuppy@126.com
通讯作者：孙恢礼（1953—），男，博导，二级研究员，研究方向：海洋生物学与资源化学。E-mail: shl@scsio.ac.cn

(来源：《渔业现代化》2010.37(1):29-32)