摘要
通过测定4种肉样品(猪肉、牛肉、羊肉及虾)的挥发性盐基氮(TVB-N)、细菌总数、pH值和感官评分等指标数据,运用支持向量机方法对以上数据进行综合训练得到数学模型,并对SVM模型参数采用粒子群优化算法进行优化,拟实现肉品新鲜度的快速准确分类.结果表明:仅采用某一项理化指标对肉品新鲜度进行判定误判率较高,而采用默认参数条件下的以RBF为核函数的SVM模型能一定程度上提高判别准确率,但利用PSO优化的SVM模型能将肉品新鲜度判别准确率提高到100%,且模型还具有极好的稳定性.
TVB-N content,total bacterial count,pH value and sensory scores of four meat samples of pork,beef,mutton and shrimp were determined.According to support vector machine(SVM) method,the experimental data were trained to optimize the model parameters by particle swarm optimization(PSO).Based on the proposed method,the rapid and correct classification of meat freshness was rea-lized.The experimental results show that it is difficult to obtain ideal classification accuracy by any single physicochemical or sensory property.The SVM model with RBF kernel function and default parameters can improve classification accuracy to some extent.The SVM model optimized by PSO can improve classification accuracy of meat freshness to 100% with high stability.
出处
《江苏大学学报(自然科学版)》
EI
CAS
北大核心
2012年第3期288-292,321,共6页
Journal of Jiangsu University:Natural Science Edition
基金
国家自然科学基金资助项目(31101348
31000780)
上海市晨光计划项目(09CG50
2008CG55)
关键词
肉品新鲜度
判别
粒子群优化算法
支持向量机
优化
meat freshness
classification
particle swarm optimization
support vector machine
optimize