【作者】 苏晓燕；

【导师】 邓勇；

【作者基本信息】 上海交通大学 ， 测试计量技术及仪器， 2011， 硕士

【摘要】 农业预警是农业风险管理的首要环节。目前,对农业进行风险预警主要沿用一般的风险预警方法,多是从某一角度作为切入点进行预警分析,尚未针对农业系统的多因素特点建立基于信息融合的分析平台。因此,建立一个有效融合多属性信息的智能化预警平台对我国农业发展具有重要的现实意义和理论价值。本文在农业部智能化农业预警技术重点开放实验室2010年开放性课题(课题编号:2010-DAEW-01)的支持下开展研究,针对农业系统存在的复杂性、模糊性、随机性、经验性和条件性等特点,建立多属性信息融合决策的农业预警模型,并开发农业预警系统,具体的研究内容和创新性成果如下:(1)基于等级全息模型建模(HHM)方法构建农业风险预警系统的指标体系。(2)根据各个指标因素对农业系统影响的程度不同,本文采用层次分析法(AHP)对各属性指标进行重要性分配,即权重确定。(3)采用D-S证据理论对多属性指标进行数据处理,避免了模糊风险分析法难以处理定量指标数据且具有很强主观性的缺陷。本文提出了将多属性数据转化为基本概率指派函数(BPA)的办法,使之成为适合于应用D-S证据理论处理的形式。该数据转化过程实质上是一种数据无量纲化的过程,它为定量定性信息融合打下基础。(4)在对比国内外关于D-S证据理论中合成规则的主要研究成果的基础上,采用基于权重系数的修改证据源的冲突信息融合方法。该方法能够较好的解决冲突证据的融合问题,收敛速度较快。(5)基于BPA函数的转换模型对农业风险进行决策。(6)基于IEDSS的设计理念,采用MatlabR2009b进行编程,并利用GUI进行界面的制作,实现了农业预警系统软件的开发。更多还原

【Abstract】 Early-warning for agricultural acts as the first link of agricultural risk management. Sofar the early warning in terms of agricultural risk has mainly adopts the leftover methodof ordinary risk pre-warning method, making a breakthrough from one single aspect in mostcases, however, failed to combine with the multi-factor features in agriculture system to buildmodels based on data fusion analyzing platform. Therefore, it is of great theoretical valueand practical significance for agriculture development to establish intelligent early warningplatform capable of effective multi-information fusion.The thesis, supported by the Open Project Program of Key Laboratory of Digital Agri-cultural Early-warning Technology (No. 2010-DAEW-01), aims to do research on differ-ent uncertainty puzzles in agricultural system, establish a multi-property information fusiondecision-making model and then develops an agriculture early warning system. The early-warning model is based on the HHM method, Analytic Hierarchy Process (AHP) methodcombined with D-S evidence theory. The research contents and creative achievements arelisted as follows:(1) In accordance with the complexity of agriculture system, we set up a indicatorsystem in view of agricultural risk early warning based on Hierarchy Holographic Mod-eling(HHM).(2) According to the distinct effect each factor has on the agricultural system, this articleadopts the Analytic Hierarchy Process (AHP) method to distribute importance to indicators,as also can be said weight confirming.(3) To avoid the subjectivity and the difficulty in processing the quantitative index inthe fuzzy risk analysis method, we use D-S theory to process the multi-property data. Thisessay puts forward a method of converting property data into basic probability assignment(BPA), rendering it the right form for D-S theory. The data conversion, in fact, is exceptingfor the effects of dimensions of quantitative and qualitative information.(4) In overall consideration of various factors of in?uence on agricultural system, thispaper introduces a collision information fusion of modified evidence source based on weightcoefficient, by comparing main advantages of D-S theory’s combination rule, the method used in this paper is domestic and foreign. This method can well solve the problem ofcollision evidence fusion in a rather high convergence speed.(5) Establish a BPA function transformation model to evaluate agricultural risk.(6) In this paper, we develop the software of agriculture early warning system basedon IEDSS design notion, under Matlab2009b programming and GUI interface designingenvironment.更多还原