基于最小诊断集的贝叶斯网络诊断模型研究

为了克服基于模型的诊断方法不能确定诊断最优解的缺陷,提出以基于模型的诊断所生成的最小诊断集构建贝叶斯网络的诊断方法,它能通过贝叶斯网络确定候选诊断解的发生概率,进而确定系统最有可能的诊断解。并针对当最小诊断解较多时,构建的贝叶斯网络将非常复杂,不易诊断的问题,提出了超件的概念,并将超件应用于贝叶斯网络构建,以简化构建的贝叶斯网络,提高诊断效率。最后以某电路为例进行诊断分析。

基于一致性的最小正常诊断及其应用

本文给出了基于一致性的最小正常诊断的概念，并将它与基于一致性的最小反常诊断加以区别．证明了老将系统描述限制在故障理论或在故障理论中扩展有限的正常行为知识，那么使用最小正常诊断能够刻画基于一致性的诊断空间，即最小正常诊断假设成立．本文还指出：使用最小正常诊断可以缩小只针对正常行为模型的最小反常诊断所产生的诊断空间，帮助我们找到真正的故障．文中最后还指出了最小正常诊断所适合的诊断任务及诊断领域．

基于组件逻辑结构关系的故障诊断方法

提出一种利用组件逻辑结构关系直接求解所有最小诊断的方法。该方法将系统的最小诊断求解分解为离线建立规则和在线确定诊断两个步骤。方法的本质属于溯因诊断。以Poly-box系统为例验证了方法的正确性,同时对方法的诊断解的完备性、可采纳性、重用性进行了分析说明。针对方法中的规则建立效率低下的问题,提出了两点改进措施,并仍以Poly-box系统为例进行了对比验证,结果表明,改进的方法提升了诊断规则的建立效率。

面向需求的应用本体不一致诊断筛选方法研究

针对现有本体诊断工具返回最小诊断数量多、用户无法科学取舍的问题,首先分析现有筛选诊断方法存在问题,然后结合应用本体及其需求特点,探讨将原本相互孤立的、分别应用在不同诊断工具中的公理可信度筛选诊断法、信息熵筛选诊断法之核心技术进行组合、优化的思路与方法。在此基础上,提出一个综合性的启发式筛选诊断法,阐述其基本思路与操作步骤,并以典型实验本体为例进一步验证方法的有效性。

基于改进LS-SVM的异步电机转子故障诊断

为了提高异步电动机转子故障的诊断精度,给出了一种基于改进最小二乘支持向量机(LS-SVM)的多故障分类算法。首先运用FFT处理电机的定子电流信号得到信号频谱图,从中提取故障特征向量;然后将特征向量送入改进算法进行故障诊断时,在原有多分类算法的基础上引入层次分析法确定故障类别的权重,根据权重值确定故障的诊断顺序,依次进行故障分类。实验表明,改进算法用于故障诊断节省了诊断时间,提高了诊断精度,具有很好的推广前景。

离散事件系统最小故障诊断基的求解与应用

针对离散事件系统(DESs)的最小故障故障诊断基(MDB)求解问题,提出了一种基于事件集树的求解算法.首先在诊断器的基础上构建了一种新型测试自动机,并求得离散事件系统可诊断性的充分必要条件,实现了对D ESs的故障诊断.在此基础上,提出了通过构造事件集树求解离散事件系统MDB的算法,并对算法的复杂度进行了分析.最后,通过网络系统中关键网路的选取实例对算法进行了分析.与现有算法相比,所提算法不仅复杂度更低,而且可用于在线诊断和离线诊断,适用性更广.

不一致术语集最小基数诊断的分支限界

针对不一致术语集的最小基数诊断,提出了一种分支限界的方法,该方法根据不相交极小不一致保持子术语集(MIPS)覆盖来定义诊断的一个下界,并且使用分支技术将大问题分解为多个可并行执行的子问题。实验结果表明,该方法能够快速有效地计算出一个最小基数诊断。

Dynamic change and diagnosis of physical, chemical and biological properties in bauxite residue disposal areas

Vegetation encroachment occurred in bauxite residue disposal area(BRDA)following natural weathering processes,whilst the typical indicators of soil formation are still uncertain.Residue samples were collected from the BRDA in Central China,and related physical,chemical and biological indicators of bauxite residue with different storage years were determined.The indicators of soil formation in bauxite residue were selected using principal component analysis,factor analysis,and comprehensive evaluation to establish soil quality diagnostic index model on disposal areas.Following natural weathering processes,the texture of bauxite residue changed from silty loam to sandy loam.The pH and EC decreased,whilst porosity,nutrient element content and microbial biomass increased.The identified minimum data set(MDS)included available phosphorus(AP),moisture content(MC),C/N,sand content,total nitrogen(TN),microbial biomass carbon(MBC),and pH.The soil quality index of bauxite residue increased,and the relative soil quality index decreased from 1.89 to 0.15,which indicated that natural weathering had a significant effect on improveing the quality of bauxite residue and forming a new soil-like matrix.The diagnostic model of bauxite residue was established to provide data support for the regeneration on disposal area.

Fault diagnosis using a probability least squares support vector classification machine

Coal mines require various kinds of machinery. The fault diagnosis of this equipment has a great impact on mine production. The problem of incorrect classification of noisy data by traditional support vector machines is addressed by a proposed Probability Least Squares Support Vector Classification Machine (PLSSVCM). Samples that cannot be definitely determined as belonging to one class will be assigned to a class by the PLSSVCM based on a probability value. This gives the classification results both a qualitative explanation and a quantitative evaluation. Simulation results of a fault diagnosis show that the correct rate of the PLSSVCM is 100%. Even though samples are noisy, the PLSSVCM still can effectively realize multi-class fault diagnosis of a roller bearing. The generalization property of the PLSSVCM is better than that of a neural network and a LSSVCM.

Diagnosis of Soil Nutrient Constraints in Small-Scale Groundnut （Arachis hyopaea L,） Production Systems of Western Kenya Using Infrared Spectroscopy

Wise decision-making on resource allocation and intervention targeting for soil management cannot rely solely on trial and error methods and field observations used by small-scale farmers： cost-effective soil fertility survey methods are needed. This study aimed to test the applicability of infrared spectroscopy （IR） as a diagnostic screening tool for making soil fertility recommendations in small-scale production systems. Soil fertility survey of 150 small-scale groundnut farms in western Kenya was conducted using a spatially stratified random sampling strategy. Soil properties examined were pH in water （pHw）, total carbon （C）, total nitrogen （N）, extractable phosphorus （P）, exchangeable potassium （K）, calcium （Ca）, magnesium （Mg） and texture. These properties were calibrated to mid-infrared （MIR） diffuse reflectance using partial least square regression （PLSR）. Cross-validated coefficient of determination （r2） values obtained from calibration models were 〉 0.80 for all properties, except P and K with 0.66 and 0.50 respectively. Soil nutritional deficiencies were evaluated using critical nutrient limits based on IR predictions and composite soil fertility indices （SFIs） developed from the soil properties using principal component analysis. The SFIs were calibrated to MIR soil spectral reflectance with cross-validated r： values 〉 0.80. The survey showed that 56% of the groundnut farms had severe soil nutrient constraints for production, especially exchangeable Ca, available P and organic matter. IR can provide a robust tool for farm soil fertility assessment and recommendation systems when backed up by conventional reference analyses. However, further work is required to test direct calibration of crop responses to spectral indicators and to improve prediction of extractable P and K tests.

基于模型诊断的发展与展望

在智能故障诊断领域中,基于模型诊断方法已经成为研究者们主要研究的对象。但是,其计算的难解性、诊断结果的数目的庞大,对实际问题没有统一、寻求适当的模型表示方法是基于模型智能故障诊断必须面临的瓶颈问题。针对国内外专家学者的研究成就和现状,本文对基于模型诊断方面的主要研究成果进行了简要地介绍,对各种方法的优点和不足进行分析,并对进一步的研究提出了一些建议。