关于粗糙模糊支持向量机的探索

基于粗糙集理论和模糊支持向量机思想,提出了一种新的支持向量机模型-粗糙模糊支持向量机(RFSVM).采用UCI机器学习数据库中的数据,对比实验,RFSVM比传统支持向量机(SVM)和模糊支持向量机(FSVM)都有明显高的测试精度.     

关于粗糙模糊支持向量机的探索

基于粗糙集理论和模糊支持向量机思想,提出了一种新的支持向量机模型--粗糙模糊支持向量机(RFSVM).采用UCI机器学习数据库中的数据,对比实验,RFSVM比传统支持向量机(SVM)和模糊支持向量机(FSVM)都有明显高的测试精度.     

一种改进的证据理论融合算法

[针对D-S证据理论无法解决证据冲突和基蝴率分配函数的微小扰动使融合结果急剧变化等问题，提出了一种改进的证据理论融合算法。该方法首先求出了证据距离得出相应的证据冲突，然后给出描述证据间相互支持度的模糊矩阵，并依次来研究不同情况对融合数据的影响，进而说明改进的融合算法的有效应。     

一种考虑传感器精度的数据自适应加权融合算法

为增强多传感器测量数据融合效果,在综合考虑传感器初始精度与实际测量精度的基础上,提出了一种改进的自适应加权融合算法。将证据理论中的修正证据距离引入传感器实际测量数据间距离计算,基于计算得出的测量数据间距离生成各数据融合时的测量权重值。当传感器精度已知或者能够计算得出时,将基于传感器精度生成的固定权重与测量权重相综合,生成最终权重;当传感器精度未知且无法计算得出时,将测量权重作为最终权重。基于多种典型算例对所提融合算法进行验证,结果表明所提算法融合效果较好,具有一定的理论意义和较好的工程实用价值。     

基于支持向量机技术的粮虫图像分类技术研究

介绍了支持向量机的基本思想，提出了一个基于支持向量机的粮虫模式识别系统。该系统先对粮虫图像进行小波边缘提取，根据灰度共生矩阵和局部统计方法提取小波分割后的图像纹理特征。最后利用支持向量机对粮虫图像进行分类。     

多源置信综合的雷达与ESM协同跟踪方法

跟踪空间邻近目标时,仅依靠运动学信息不足以实现可靠的数据关联,而基于动目标指示器（Moving Target Indicator,MTI）雷达和电子支援措施（Electronic Support Measurement,ESM）的多源异类传感器数据融合可以通过提升数据关联性能达到改善跟踪性能的目的。通过构建基于五种成比例再分配规则（Five Proportional Conflict Redistribution Rules,PCR5）置信度量的数据关联策略,将目标运动学信息和属性信息结合做多特征推理,解决异类传感器数据的不确定性和不一致性;利用Dempster-Shafer（DS）证据理论方法进行属性融合更新,完成属性信息在时间序列上的相干积累,实现空间邻近目标的可靠跟踪。该方法从数据关联和状态估计两方面联合进行改进,通过引入属性信息提升数据关联的正确性,从而提升跟踪性能,实现多源异类信息下的协同跟踪。仿真表明,相比于仅雷达跟踪、雷达和ESM序惯跟踪等方案,该方法可有效提升跟踪精度和关联性能。     

支持向量机在模式识别中的应用

针对传统神经网络存在网络结构难于确定、过学习以及局部极小等问题，研究了基于支持向量机（SVM）的模式识别问题。通过对棋盘这种典型非线性二值问题的分类研究，分析了支持向量机的分类与泛化能力。支持向量机在分类和泛化能力方面远远优于传统神经网络。最后将支持向量机用于对两类飞机目标的分类识别，通过多组蒙特卡罗试验，获得了较好的识别结果。支持向量机在目标识别中有巨大潜力和广阔前景。     

一种LAMBDA整周模糊度解算正确性的评估算法

针对基于经典假设检验理论的模糊度正确性评估算法的缺陷，提出了一种依据概率指标的LAMBDA（Least Square Ambiguity Decorrelation Adjustment）整周模糊度解算的正确性评估算法。该算法能够有效解决实时动态差分定位中模糊度解算值的可信度问题，降低系统完好性风险。通过对跑车实验中GPS差分定位数据的直接处理及引入多径误差后处理，该评估算法可以在置信水平为99.5%的条件下实时有效地评估模糊度解算值的正确性。     

基于核Fisher鉴别分析的网络入侵检测算法

提出了一种将核Fisher鉴别分析特征抽取与多分类支持向量机算法结合的网络入侵检测技术,扩展了二分类支持向量机．利用经过核Fisher鉴别分析特征抽取后的训练数据构造优化的决策树,从而实现支持向量机的多分类。实验结果表明该算法能够提高检测正确率,同时降低训练时间,取得了良好的效果。     

基于证据源和组合规则双修正的证据合成

针对冲突证据合成时传统证据合成和单一修正改进方法存在的问题，提出了新的证据源预处理和组合规则双修正的证据合成方法。利用夹角余弦理论，计算证据向量距离后转化得到权重系数，依据权重系数对证据进行重新概率分配，完成证据源修正；之后用修正的动态合成规则，进行修正证据的合成;同时提出了算法分析模型。实例结果验证了算法在证据合成时具有最优结果、较强通用性和稳定性，可应用于信息融合、不确定信息决策等领域。     

Determining the causal relationship among the research streams of the operations management course

Due to the rapid changes in technology, there have been dramatic changes in manufacturing systems and techniques in almost all industries. In this sense, higher education institutions have become more important as they play a key role in ensuring the labor force in the industry. In the present study, the research streams for operations management course were established, then fuzzy total interpretive structural modelling technique was employed to analyze the causal relations and influences among the factors. The purpose of the study was to determine the relationships among the research streams of operations management course.     

A linguistic decision process in group decision making

Assuming a set of linguistic preferences representing the preferences of the individuals, a linguistic choice process is presented. This is developed using the concept of fuzzy majority for deriving a collective linguistic preference, and the concept of nondominated alternatives for deriving the selected alternatives in the linguistic choice process. The fuzzy majorities are equated with fuzzy linguistic quantifiers. The collective linguistic preference is derived by means of a linguistic ordered weighted averaging operator whose weights are defined using a fuzzy linguistic quantifier. In order to obtain the nondominated alternatives, we present a novel reformulation of Orlovski's nondominance degree under linguistic information.     

模糊熵在固有风险评估中的应用探讨

对审计风险实施有效管理的前提是对风险水平的评估,它决定审计工作的成败。针对当前审计实践中主要依靠注册会计师的主观职业判断进行风险评估的现状,必须以固有风险为研究对象,来探讨模糊熵和模糊综合评价法在固有风险评估的应用,以缩小主观判断和实际水平之间的差异,更有效地保证审计质量和审计效率的提高。     

An Integrated Group Decision-Making Method Dealing with Fuzzy Preferences for Alternatives and Individual Judgments for Selection Criteria

Organizations often require decisions to be made by a group, and decision makers often have fuzzy preferences for alternatives and individual judgments when attempting to reach an optimal solution. In order to deal with the fuzziness of preference of decision makers, this paper proposes an integrated fuzzy group decision-making method. This method allows group members to express fuzzy preferences for alternatives and individual judgments for solution selection criteria. It also allowed for the weighting of group members. The method then aggregates these elements into a compromise group decision which is the most acceptable for the group as a whole. This method has been implemented and tested. An example is presented to illustrate the method.     

基于模糊积分的新产品开发合作伙伴选择

新产品开发合作伙伴选择需要考虑的因素很多,既有定性的也有定量的。对于定性指标运用三角模糊数将其量化,并引入核心企业的风险态度因子,通过映射将决策信息转化为点值决策信息;同时利用熵权的求法来确定模糊密度。在此基础上建立基于模糊积分的优选模型,从而选出最满意的合作伙伴,并通过实例进行有效分析。     

含有模糊权值的模糊质量综合评判

针对权值的不确定性,结合L-R模糊数的近似算法,用海明距离测度工具来度量不同评语集之间的差异,提出一种新的模糊质量评判方法,该方法能够更有效处理质量综合评判中的不确定性的非量化信息。用广义的归一化方法对模糊权值和模糊评判值归一化处理,使评判结果能够落在给定的区间内。此外,含有模糊权值的模糊质量综合评判方法,可以作为基本方法应用于质量工程的研究中。     

模糊群决策方法在中间商选择中的应用研究

企业在选择中间商时往往面临着候选伙伴评价指标信息获取的困难,为了使决策科学合理,采用群决策方法集成各领域不同专家的信息,通过语言变量把专家的主观判断转化为对应模糊数,建立相应的数学模型,并计算出用模糊数表示的候选者综合评价值。     

人员选聘的直觉模糊复合决策模式

在探讨了人员选聘的意义和标准的基础上,根据人员选聘标准具有模糊性,以及复合决策中不同岗位对人员的个性能力有不同要求,不同人员的个性能力也各不相同的特点,给出了一种直觉模糊复合决策模式,并结合模拟算例说明了该模型的应用,具有较强的实用性和信度。     

模糊多属性决策法在配送中心选址中的应用

依据配送中心选址的主要因素,从基础设施、自然环境、社会效益、费用以及服务水平等几个方面建立多级评判指标体系,并提供两种模糊多属性决策法对备选配送中心进行优选和排序。最后一个算例方法客观反映实际情况,具有可行性。     

A risk-based decision support framework for selection of appropriate safety measure system for underground coal mines

In the context of underground coal mining industry, the increased economic issues regarding implementation of additional safety measure systems, along with growing public awareness to ensure high level of workers safety, have put great pressure on the managers towards finding the best solution to ensure safe as well as economically viable alternative selection. Risk-based decision support system plays an important role in finding such solutions amongst candidate alternatives with respect to multiple decision criteria. Therefore, in this paper, a unified risk-based decision-making methodology has been proposed for selecting an appropriate safety measure system in relation to an underground coal mining industry with respect to multiple risk criteria such as financial risk, operating risk, and maintenance risk. The proposed methodology uses interval-valued fuzzy set theory for modelling vagueness and subjectivity in the estimates of fuzzy risk ratings for making appropriate decision. The methodology is based on the aggregative fuzzy risk analysis and multi-criteria decision making. The selection decisions are made within the context of understanding the total integrated risk that is likely to incur while adapting the particular safety system alternative. Effectiveness of the proposed methodology has been validated through a real-time case study. The result in the context of final priority ranking is seemed fairly consistent.