【作者】 徐峰；

【导师】 傅泽田；

【作者基本信息】 中国农业大学 ， 管理科学与工程， 2005， 硕士

【摘要】 本文选题来源于863计划现代农业技术“网络化农业病虫害远程诊断平台与示范”和”数字畜牧业精细养殖平台技术研究与示范”课题。本文从我国生猪养殖实践中的难题——猪病诊断出发,针对现有猪病诊断专家系统的不足之处,提出了全新的解决方案,并对构建猪病诊断专家系统的各项技术做了全面的论述,进而提炼出了一些适合猪病诊断特点的研制猪病诊断专家系统的方法。 本文首先通过领域知识的研究,分析了生猪的生物学和行为学特性,在此基础上给出了猪病诊断的标准症状表现,同时,在领域专家的协助下,总结了专家的诊断推理思维;其次,根据已经获得的领域基础知识,进一步描述了有关知识的获取、分类、标准化、症状分期、权重和基本概率分配的过程;最后归总以上的研究成果,创造性地把证据理论和集对分析理论应用于同一个诊断推理中,构建了”症状—疾病—症状”正反双向推理模型。由于诊断的侧重点不同,在猪病诊断推理的不同阶段,本文分别采用了不同的诊断指标,在“诊断”阶段,主要运用证据理论的方法,对证据(症状)对假设(疾病)的基本概率进行了分配,并在此基础上实现推理运算;而在“验证”阶段,则主要根据权重和模糊度指标求解所选症状集合与标准症状集合的相似度。 研究结果表明,在“诊断”阶段采用证据理论有利于并发疾病的诊断,而在“验证”阶段采用集对分析则可优化集合对比分析的角度。两者相结合可有效解决目前猪病诊断专家系统存在的诊断模糊性问题,而症状分期则显著提高了猪病诊断的精准度。本系统能诊断54种猪病及127组并发猪病集合,可基本满足各类用户的需要。更多还原

【Abstract】 In order to solve a difficult question-pig disease diagnosis which is subsistent in the pigindustry of our country, the means which can be used to guide developing pig disease diagnosis expert system from the point of technical view was brought forward in the dissertation After analyzing theoretically conceptions, the diagnostic knowledge and the diagnosis reasoning process, a double deck positive and negative network model including disease and symptom factors was constructed in the article In the symptom-disease deck, on the one hand, aim at the time characteristic of symptom, the symptom is divided into two stages: the prophase symptom and the anaphase symptom Furthermore, every symptom is endowed probability by the process of advance disposal On the other hand, the knowledge base based on the basic probability assignment and the knowledge base based on posterior diagnostic probability were set up, and the D/S theory was adopted in uncertainty reasoning model of pig disease diagnosis In the disease-symptom deck, according to the relation between the set users selected and the standard set, following the same, different and reverse principle, the diagnosis index which is based on the correlative degree and the fuzzy degree was formed with the way of Set Pair Analysis ,The system has over 325 symptoms, 54 pig diseases and 316 pictures and graphics for different types of diseases and symptoms It can diagnose 127 types of common disease sets of pigs, so it can meet the demand of users basically This article relies on Chinese National Funded High Science and Technology Research Project (863 project) under Grand No.2003AA209050更多还原