奶牛行为特征识别方法的研究与实现——基于支持向量机  

Research and Implementation of Cattle Behavior Character Recognition Method——Based on Support Vector Machine

作　　者：侯云涛[1] 蔡晓华[1] 吴泽全[1] 东忠阁 Hou Yuntao;Cai Xiaohua;Wu Zequan;Dong Zhongge(Scientific Research Institute of Agricultural Mechanical Engineering in Heilongjiang, Harbin 150081, China;Harbin Bona Technology Co Ltd,Harbin 150081, China)

机构地区：[1]黑龙江省农业机械工程科学研究院,哈尔滨150081 [2]哈尔滨博纳科技有限公司,哈尔滨150081

出　　处：《农机化研究》2018年第8期36-41,共6页Journal of Agricultural Mechanization Research

基　　金：黑龙江省属科研院所基本科研业务费项目(2017-69)

摘　　要：随着我国规模化奶牛养殖场数目的快速增加,传统的依靠人工判断奶牛发情的方法在效率和准确率方面已无法满足需要。为此,设计了一种配置有加速度传感器的、基于Zig Bee无线传感技术的传感器节点,通过线圈的形式佩戴在奶牛脖子上,在不干涉其正常活动的情况下,实时监测并采集奶牛三轴运动加速度数据,并利用基于二叉决策树支持向量机(Support Vector Machine,SVM)时间序列模型判断奶牛个体的行为特征,了解和掌握奶牛个体的身体状况。测试结果表明:分类模型的整体测试结果是较好的,建立的多分类模型能够较准确地识别奶牛个体的行为特征,为及早发现奶牛病情和发情提供了有力的支持。As the number of scale dairy farms in China rapidly increasing,The efficiency and accuracy of the traditional judgment method relying on artificial judgment of cow estrus have been unable to meet the need. This study designs a configuration of the acceleration sensor based on Zig Bee sensor nodes of wireless sensor technology,through the form of coil worn in the cow neck,in the case of not to interfere in their normal activities,based on real-time monitoring and gathering cow triaxial acceleration data. Use Support Vector Machine（SVM） based on binary decision tree with time series model to judge the cow individual behavior characteristics,and preliminarily determining whether the cow is in heat.Test results show that the overall test results of classification model is good,the classification model can determine more accurately cows individual behavior characteristics,for the preliminary identification of cow estrus provides strong support.

关 键 词：奶牛 行为特征 支持向量机 发情鉴定 

分 类 号：S815.4[农业科学—畜牧学]