文献类型：会议论文

英文题名：Prediction Model of Ocean Food Microbe Growth Based on Neural Network and Its Simulation

作者：Wang Zhengxia[1,2];Xiao Laisheng[3];Li Xiaomei[2];Tian Jinling[4]

机构：[1]Guangdong Ocean Univ, Coll Law, Zhanjiang, Peoples R China;[2]Guangdong Univ Technol, Fac Comp, Guangzhou, Guangdong, Peoples R China;[3]Guangdong Ocean Univ, Network & Educ Technol Ctr, Zhanjiang, Peoples R China;[4]Guangdong Ocean Univ, Coll Food Sci & Technol, Zhanjiang, Peoples R China

会议论文集：International Conference on Computers, Communications, Control and Automation (CCCA 2011)

会议日期：FEB 20-21, 2011

会议地点：Hong Kong, PEOPLES R CHINA

语种：英文

外文关键词：neural network; microbe growth model; vibrio parahaemolyticus; ocean food; simulated crab meat

外文摘要：In processing oceanic aquatic food, the growth of microbe is influenced by many environmental factors, where temperature and salinity are two important ones. The objective of this paper is to set up a neural network-based prediction model for the growth of vibrio parahaemolyticus according to the effect of temperature and salinity to it in oceanic aquatic processed food, simulated crab meat, which provided a tool of analysis for secure production and processing of simulated crab meat. Firstly, we took a large number of experiments to study relationship between the growth of vibrio parahaemolyticus and the time in different temperature and different salinity, and acquired a lot of experimental data. Next, by applying BP neural network to accomplish analyzing and modeling for the data, two-dimensional and three-dimensional neural network-based simulation models for simulating vibrio parahaemolyticus growth were founded. Their performance was also researched by means of computer simulation technology Matlab and compared with traditional Gompertz model. Simulation results showed that the precision of two-dimensional neural network model was ten orders of magnitude better and higher than Gomperts model for MAE and MSE. For three-dimensional neural network model, we also obtained less appraisal target value of MAE and MSE. Therefore, neural network-based microbe growth prediction model has a lot of advantages including less error, higher precision and stable and reliable performance.