面向肉鸡疾病综合诊断的智能化平台及关键诊断流程研究  

Intelligent Platform and Key Diagnostic Process for Comprehensive Diagnosis of Broiler Diseases

作　　者：彭星凯 孔祥超 孙博文 陈长喜 郭永敏 向斌 张立梅 张万潮 赵光煜 杨恺斯 PENG Xingkai;KONG Xiangchao;SUN Bowen;CHEN Changxi;GUO Yongmin;XIANG Bin;ZHANG Limei;ZHANG Wanchao;ZHAO Guangyu;YANG Kaisi(College of Computer and Information Engineering,Tianjin Agricultural University,Tianjin 300384;Key Laboratory of Smart Breeding(Co-construction by Ministry and Province),Ministry of Agriculture and Rural Affairs,Tianjin 300384;College of Veterinary Medicine,Yunnan Agricultural University,Kunming,Yunnan 650201)

机构地区：[1]天津农学院计算机与信息工程学院,天津300384 [2]农业农村部智慧养殖重点实验室(部省共建),天津300384 [3]云南农业大学动物医学院,云南昆明650201

出　　处：《中国家禽》2025年第4期152-160,共9页China Poultry

基　　金：云南高原特色畜禽产业提质增效关键技术集成与示范(2022YFD1601904);财政部和农业农村部:国家现代农业产业技术体系(CARS-41);国家重点研发计划(2023YFD2000801)。

摘　　要：为解决传统肉鸡疾病诊断方法存在的信息传递效率低、诊断时效性不足以及经济与时间成本高等问题,文章提出了一种面向肉鸡疾病综合诊断的智能化平台,并对其关键诊断流程进行研究。结果显示:该平台集成了数据采集与上传、机器学习模型预测、专家诊断与审核流程、结果反馈与决策支持以及用户权限设置等模块;通过专家诊断、畜牧站审核和政府研判的融合,实现疾病的精准诊断、早期预警与高效防控;在模型性能方面,采用粒子群优化算法进行特征选择,并结合蜣螂优化算法对机器学习模型进行参数优化,提升预测性能;随机森林模型在最优特征集和参数配置下与粒子群优化算法和蜣螂优化算法的组合算法预测准确率达到94.02%,案例的综合诊断准确率为95.6%。研究表明,建立的肉鸡疾病综合诊断智能化平台在实际应用中诊断准确率较高,能够提升肉鸡养殖企业、养殖户的疾病诊断能力,减少养殖风险,为肉鸡养殖业的智能化、精准化疾病预警与防控提供新思路和技术支持。To address the challenges of traditional methods for diagnosing broiler diseases with low information transmission efficiency,limited timeliness in diagnosis,and high economic and time costs,this study proposed an intelligent platform for comprehensive diagnosis of broiler disease and conducted an in-depth investigation into its key diagnostic processes.The results showed that the platform integrated modules for data collection and uploading,machine learning model prediction,expert diagnosis and review processes,result feedback and decision support,as well as user permission settings.The integration mechanism of expert diagnosis,livestock station review and government assessments,provided precise diagnosis,early warnings and effective prevention and control of diseases.For model performance,particle swarm optimization for feature selection and dung beetle optimization for model parameter tuning were used to achieve improvements in predictive accuracy of platform.Under an optimal feature set and parameter configuration,the Random Forest model achieved a prediction accuracy of 94.02%with Particle Swarm Optimization and Ant Colony Optimization,and there was a comprehensive diagnostic accuracy of 95.6%for the case.The results indicated that the established intelligent platform for comprehensive diagnosis of broiler disease had high diagnostic accuracy and could enhance disease diagnostic capabilities for broiler farming enterprises and farmers in practical applications,and reduced farming risks and offered novel insights and technological support for intelligent and precise disease early warning and prevention in broiler industry.

关 键 词：肉鸡 疾病 综合诊断 机器学习 兽医专家 畜牧站 政府 

分 类 号：S854.4[农业科学—临床兽医学] TP29[农业科学—兽医学]