摘要
为了提高在鸡种蛋孵化早期(0~5胚龄)筛除无精蛋的准确率,试验采用透射光谱技术结合智能算法与机器学习模型进行种蛋受精信息识别。试验对采集获得的透射光谱数据进行预处理,剔除壳色波段影响,建立支持向量机(Support vector machine,SVM)种蛋受精检测模型。分别使用灰狼优化算法(Grey wolf optimizer,GWO)和麻雀搜索算法(Sparrow search algorithm,SSA)对SVM模型的c和g参数进行优化,优化后模型进行对比;采用Sine混沌映射和萤火虫扰动优化麻雀搜索算法,构建改进SSA-SVM模型。结果显示:SSA-SVM模型对孵化早期测试集预测受精准确率在孵化第4、5天最高,达99.56%;改进后的SSA-SVM模型对入孵前第0天测试集预测受精准确率达99.12%。研究表明使用改进后的SVM模型能够提高种蛋受精判别准确率,可以为生产提供参考。
In order improve the accuracy of screening unfertilized eggs in the early incubation period(0 to 5 days)of chicken eggs,transmission spectros-copy combined with intelligent algo rithms and machine learning models for the identification of fertilization information of breeding eggs were used.The test preprocessed the acquired transmission spectrum data,and then eliminated the effect of shell color band,and support vector machine(SVM)model was established to detect the fertilization information of breeding eggs.Then,the c and g parameters of the SVM model were optimized using the primitive gray wolf optimization(GWO)algorithm and the primitive sparrow search algorith m(SSA).Additionally,Sine chaotic mapping and Firefly perturbation were used to further optimize the SSA,creating an improved SSA-SVM model to enhance the accuracy of fertilized egg detection before incubation.The results showed that SSA-SVM model had the highest fertilization prediction accuracy of 99.56% on the 4th and 5th day of incubation.The improved SSA-SVM model was 99.12% accurate in predicting fertilization on the test set 0 days before hatching.The results indicated that the improved SVM model could improve the accuracy of egg fertilization discrimination and provide reference for future production.
作者
刘云飞
张晓雨
籍颖
周荣艳
陈辉
韩晓飞
LIU Yunfei;ZHANG Xiaoyu;JI Ying;ZHOU Rongyan;CHEN Hui;HAN Xiaofei(College of Information Science and Technology,Hebei Agricultural University,Baoding,Hebei 071000;Hebei Key Laboratory of Agricultural Big Data,Baoding,Hebei 071000;College of Animal Science and Technology,Hebei Agricultural University,Baoding,Hebei 071000;Key Laboratory of Broiler and Layer Breeding Facilities Engineering,Ministry of Agriculture and Rural Affairs,Baoding,Hebei 071000;Huayu Agricultural Science and Technology Co.,Ltd.,Handan,Hebei 056000;Hebei Layer Industry Research Institute,Handan,Hebei 056000)
出处
《中国家禽》
北大核心
2025年第6期153-161,共9页
China Poultry
基金
邯郸市科学技术研究与发展计划项目(22313014017)
鸡现代种业科技创新团队(21326303D)
驻冀高校研究生来石实践补贴项目。
关键词
种鸡蛋
透射光谱
无损检测
支持向量机
麻雀搜索算法
灰狼优化算法
hatching egg
transmission spectrum
non destructive testing
support vector machine
gray wolf optimization
sparrow search algorith
