径向基神经网络在猪胴体瘦肉率预测中的应用

猪胴体瘦肉率(LMP)是评价猪肉品质的重要指标之一,在生产线上快速而准确地预测出其数值并进行分级是并不可少的。目前,国内大部分厂家依然采取屠宰后人工称重测量的方法,耗时耗力,且存在相当大的误差。为此,随机抽取了116头皖北地区商品猪,选定眼肌面积、背膘厚及腿臀比作为参考数据,以Mat Lab工具箱作为研究工具,利用BP、Elman和RBF等3种不同的神经网络建立预测模型,统计后进行比较分析。实验表明:3种模型的神经网络均可用于瘦肉率预测,但RBF网络误差最小,训练速度最快,学习能力最强,最适合用于建立瘦肉率的预测模型。

Lean Meat Percentage Prediction of Pig Carcass Based on Radial Basis Function Neural Network

Lean meat percentage ( LMP) of pig carcass is one of the important indexes of pork quality evaluation .It is necessary to predict the data accurately and quickly , and complete grading on the production line .Most of the producers still use the traditional manual dissection method , which is not only time and labor consuming , but also inaccurate , to measure the LMP after slaughtering .Research of neural network in this field of LMP measurement has been conducted , but was always operated on BP neural network structure .In this study , 116 commercial pigs from the northern Anhui Province were randomly selected as the research materials , and the parameters acquired were loin eye area , back-fat thickness and ham percentage .MatLab was used to build up and train BP network , Elman network and RBF network , and to generate LMP values for further analysis .As the data shows , all three types of neural networks can be used for LMP prediction, while RBF, with the minimum error , the fastest training speed and the strongest learning ability , proves to be the most suitable model for LMP prediction .