基于证据K-NN改进算法的水声目标识别

证据分类算法已广泛应用于目标识别当中。针对传统证据K近邻算法在近邻证据组合规则上的局限,研究一种新的基于PCR5规则的证据K近邻改进算法(IEK-NN)。首先在总样本集中随机重复采样来构造多个训练子集;然后在各训练子集中,利用目标数据与其近邻的特征距离来构造基本置信指派;最后利用证据推理中的PCR5规则对近邻证据进行融合,并根据融合结果以及所建立的分类规则判断目标的类别属性。通过水声目标实测数据实验,将IEK-NN与传统的证据近邻分类算法进行对比分析,结果表明新算法能有效提高识别的准确率。     

基于C-BRB模型的发动机运行状态监测方法

针对发动机运行状态监测过程中发动机内部多个因素之间相关性与建模方法可解释性问题,提出数据驱动下C-BRB方法。该方法首先通过样本数据计算发动机内部多个因素之间的Kendall秩相关系数,并确定具体Copula模型及参数λ,实现对多个因素之间相关性的测量;然后使用置信规则库(BRB)对发动机内部多个因素建模,在BRB推理过程中,每条激活规则的综合匹配度由Copula模型对该规则中各前提属性的匹配度进行计算得到,并利用证据推理(ER)算法对所有激活规则进行融合得到输出。实例结果表明,所提方法在推理发动机传感器数据上具有较高的精度。     

基于规则推理网络的分类模型

为了缓解神经网络的"黑盒子"机制引起的算法可解释性低的问题,基于使用证据推理算法的置信规则库推理方法(以下简称RIMER)提出了一个规则推理网络模型.该模型通过RIMER中的置信规则和推理机制提高网络的可解释性.首先证明了基于证据推理的推理函数是可偏导的,保证了算法的可行性;然后,给出了规则推理网络的网络框架和学习算法,利用RIMER中的推理过程作为规则推理网络的前馈过程,以保证网络的可解释性;使用梯度下降法调整规则库中的参数以建立更合理的置信规则库,为了降低学习复杂度,提出了"伪梯度"的概念;最后,通过分类对比实验,分析了所提算法在精确度和可解释性上的优势.实验结果表明,当训练数据集规模较小时,规则推理网络的表现良好,当训练数据规模扩大时,规则推理网络也能达到令人满意的结果.     

基于粗糙集的约简算法及规则融合方法的研究

从不同的角度分析了属性约简的两种重要方法:区分矩阵法和基于属性重要性.根据数据集的实际情况提出了一种基于粗糙集的区分矩阵和属性重要性相结合的启发式算法,并获得了属性约简集.在约简集的基础上分析了静态决策推理规则及算法.在相容决策系统中利用集合向量包含度构造了规则融合的方法,从而得到动态条件规则的极大近似决策值.在知识满足分类质量要求的前提下,根据规则融合方法,对任意给定的样本知识可以判别知识的实际归属类.     

基于证据距离的改进DS/AHP 多属性群决策方法

证据推理/层次分析(DS/AHP)方法采用了AHP法的层次结构模型和证据理论的分析过程,为不确定多属性决策问题的解决提供了新思路,但其在构造知识矩阵中用0代表残缺信息是不合理的.鉴于此,对DS/AHP方法进行了改进,并将改进后的方法拓展到群决策中,研究了专家群体权向量的确定方法.具体地,引入证据距离的概念,通过计算专家证据的综合距离来对专家赋权,体现了群决策中的多数人规则.     

融合机器学习与知识推理的可解释性框架

针对基于规则的可解释性模型可能出现的规则无法反映模型真实决策情况的问题, 提出了一种融合机器学习和知识推理两种途径的可解释性框架. 框架演进目标特征结果和推理结果, 在二者相同且都较为可靠的情况下实现可解释性. 目标特征结果通过机器学习模型直接得到, 推理结果通过子特征分类结果结合规则进行知识推理得到, 两个结果是否可...     

分层级故障诊断策略与规则的优先权排序方法

针对某些复杂设备很难有效地创建基于元器件层次分解的诊断树的特点,提出基于故障类别信息的分层级故障诊断推理策略,给出相应的规则存储模式并引入规则优先权;然后提出基于模糊多属性群决策理论的规则优先权排序方法;最后给出该诊断推理策略的可信度的传递算法和具体推理流程;应用实例和分析表明,所提出的方法在冲突消解方面具有很好的效果,可有效提升诊断效率.     

基于分层式证据推理的信息融合故障诊断方法

针对基于信息融合的故障诊断方法中,诊断证据的精细化获取问题和在线诊断信息量受限问题,提出分层式的证据推理(ER)诊断方法.在诊断证据获取过程中,给出故障特征参考值投点方法,按比例求取特征样本点对相邻参考值的相似度,生成点值型参考证据矩阵(REM)和在线故障特征样本的诊断证据,实现了诊断信息的精细化提取;在证据融合过程中,设计分层式ER融合模型.第1层融合中利用k-NN算法找到在线样本的近邻历史样本,然后利用ER规则实现在线样本与近邻历史样本对应证据的融合.在第2层融合中,将多个特征源提供的第1层融合结果再次融合,并根据两层融合所获证据进行故障决策;此外,在分层融合模型中,根据证据之间的欧氏距离构造目标函数及相应的证据重要性权重优化方法.最后,在多功能电机转子试验台上实施了故障诊断实验,与已有单层ER模型诊断结果进行比较,说明所提方法通过提升诊断证据的精确性、增加历史样本扩充诊断信息量,能够有效提升确诊率.     

一种基于自适应证据推理规则的集成学习方法

当前集成学习中的结合策略难以兼顾各个基学习器之间的信息和模型的可解释性。使用证据推理（evidential reasoning,ER）规则作为结合策略,将各个基学习器结果作为证据参与融合,可以较好地解决以上问题。但传统ER规则的证据参数是单一的,对不同的基学习器模型使用相同的证据参数显然是不合理的。为此,提出一种基于自适应证据推理（adaptive-evidential reasoning,A-ER）规则的集成学习方法,该方法在每次证据融合前对证据的类别进行判断,针对不同的证据类别自适应分配不同的证据参数。通过不同的分类案例表明,该方法与案例中其他方法相比具有更高的分类精度,证明了该方法使证据参数设置更加合理且具有更好的可解释性和泛化能力。     

面向大型数据表的粗分析方法

该文运用粗糙集理论,提出一种提取大型数据表中决策规则的方法。首先根据条件属性及属性值的权重将数据表分解成多个子表,再利用粗分析方法分别对各子表简约,综合每个子表归纳出的具有一定置信度的子规则形成规则集,做出推理和决策。     

基于改进模糊熵和证据推理的多属性决策方法

针对证据推理方法框架下属性权重难以获取的问题,提出一种基于改进模糊熵和证据推理的多属性决策方法。首先,定义证据推理信度决策矩阵框架下的三角函数模糊熵公式,并证明了其满足熵的四个公理化定义。其次,所提方法能够同时处理属性权重完全未知和属性权重信息部分已知两种情况：当属性权重完全未知时,基于信度框架下的改进模糊熵和熵权法的基本思想计算属性权重;当属性权重信息部分已知时,定义加权模糊熵,建立期望模糊熵最小的线性规划模型求解最优属性权重。最后,利用证据推理算法融合方案属性值,结合期望效用理论得到方案排序结果。通过实例计算,并与传统模糊熵计算方法进行比较分析,验证了所提方法能够更加充分地反映原始决策信息,更具客观性和一般性。     

Fuzzy multiattribute group decision making based on intuitionistic fuzzy sets and evidential reasoning methodology

In this paper, we propose a new fuzzy multiattribute group decision making method based on intuitionistic fuzzy sets and the evidential reasoning methodology. First, the proposed method uses the evidential reasoning methodology to aggregate each decision maker’s decision matrix and the weights of the attributes to get the aggregated decision matrix of each decision maker. Then, it uses the obtained aggregated decision matrices of the experts, the weights of the experts and the evidential reasoning methodology to get the aggregated intuitionistic fuzzy value of each alternative. Finally, it calculates the transformed value of the obtained intuitionistic fuzzy value of each alternative. The smaller the transformed value, the better the preference order of the alternative. The proposed method can overcome the drawbacks of the existing methods for fuzzy multiattribute group decision making in intuitionistic fuzzy environments.     

基于证据推理和第3 代前景理论的不确定性多属性决策方法

针对多状态不确定性多属性决策问题, 建立基于证据推理和第3 代前景理论的决策方法. 首先, 给出不确定性知识表示方法—– 确定因子结构及其构造方法; 然后, 将第3 代前景理论构造价值函数和确定权重函数引入决策方法中, 得到每个方案在各属性下的前景价值; 进一步, 根据证据推理方法对前景价值进行信息融合得到各方案的合成前景价值, 并依据合成前景价值对方案进行排序; 最后, 通过算例验证了所提出方法的可行性和有效性.

     

MADM method based on prospect theory and evidential reasoning approach with unknown attribute weights under intuitionistic fuzzy environment

This paper proposes an intuitionistic fuzzy decision method based on prospect theory and the evidential reasoning approach, aiming at analyzing multi-attribute decision making problems in which the criteria values are intuitionistic fuzzy numbers and the information of attributes weights is unknown. Firstly, the measures of entropy and cross entropy are defined for intuitionistic fuzzy sets by taking into consideration the preference of decision maker towards hesitancy degree. Secondly, combined with bounded rationality, the prospect decision matrix is calculated in the light of prospect theory and intuitionistic fuzzy distance. Thirdly, the correlational analyses are conducted between the attribute weights and three indicators which are entropy, cross entropy and prospect value, and optimization models for identifying attribute weights are built under the circumstances that the weights are incomplete and unknown. Finally, in order to avoid the loss of decision making information, the evidential reasoning approach is applied to the calculation of comprehensive prospective values for all alternatives. Following the value calculation, the ranking and the optimal alternative are determined based on the comprehensive prospective values. Illustrating examples demonstrate that the proposed method is reasonable and feasible.     

基于证据推理和VIKOR的二维语言多准则群决策方法

针对专家给出二维语言评价信息的多准则群决策问题,提出基于证据推理和VIKOR的决策方法。首先, 从专家的心理认知和二维语言评价信息的语义出发,设定函数将二维语言信息映射为信度结构;接着基于所提出的广义信度结构,及证据的Pignistic概率距离,定义广义信度结构的距离;最后将专家给出的二维语言决策矩阵转化为信度决策矩阵,用证据推理算子集结为综合信度决策矩阵,并利用VIKOR方法对其求解,获取方案排序。实例分析表明了所提出方法的有效性和实用性。     

TOPSIS with belief structure for group belief multiple criteria decision making

The technique for order performance by similarity to ideal solution(TOPSIS)is one of the major techniques in dealing with multiple criteria decision making(MCDM)problems, and the belief structure(BS)model has been used successfully for uncertain MCDM with incompleteness, impreciseness or ignorance. In this paper, the TOPSIS method with BS model is proposed to solve group belief MCDM problems. Firstly, the group belief MCDM problem is structured as a belief decision matrix in which the judgments of each decision maker are described as BS models, and then the evidential reasoning approach is used for aggregating the multiple decision makers' judgments. Subsequently, the positive and negative ideal belief solutions are defined with the principle of TOPSIS. To measure the separation from ideal solutions, the concept and algorithm of belief distance measure are defined, which can be used for comparing the difference between BS models. Finally, the relative closeness and ranking index are calculated for ranking the alternatives. A numerical example is given to illustrate the proposed method.     

信息不完全下多属性群决策的证据合成方法

针对不完全信息下多属性群决策问题,分析决策者判断信息的可靠性对群决策结果的影响,提出一种基于相对可靠度的证据合成方法。首先对定量和定性属性值进行归一化处理;然后分析决策者判断信息的相对可靠度,运用Dempster合成法则对所有焦元的基本概率分配值进行计算与合成,并给出证据推理方法的主要步骤;最后给出了一个算例。     

Data Association in Multi-target Tracking Using Belief Function

In this paper, we present a new method for data association in multi-target tracking. The representation and the fusion of the information in our method are based on the use of belief function. The proposal generates the basic belief mass assignment using a modified Mahalanobis distance. While the decision making process is based on the extension of the frame of hypotheses. Our method has been tested for a nearly constant velocity target and compared with both the nearest neighbor filter and the joint probabilistic data associations filter in highly ambiguous cases. The results demonstrate the feasibility of the proposal and show improved performance compared to the aforementioned alternative commonly used methods.     

基于分类优化的离散证据推理

离散信息在专家系统、模式识别、决策分析等领域普遍存在, 为了解决这类信息融合问题, 提出一种离散证据推理方法. 首先, 将每个离散证据拆分成一类单点值证据; 然后, 以冲突最小化为目标修正类内证据, 并采用证据推理进行组合; 最后, 以同样的方法对类间证据进行修正与组合. 所提出方法不仅可以解决离散证据的内外部冲突问题, 而且能够克服运算量过大的问题. 算例分析表明了所提出的方法是合理且有效的.

     

Reliability measure assignment to sonar for robust target differentiation

This article addresses the use of evidential reasoning and majority voting in multi-sensor decision making for target differentiation using sonar sensors. Classification of target primitives which constitute the basic building blocks of typical surfaces in uncluttered robot environments has been considered. Multiple sonar sensors placed at geographically different sensing sites make decisions about the target type based on their measurement patterns. Their decisions are combined to reach a group decision through Dempster-Shafer evidential reasoning and majority voting. The sensing nodes view the targets at different ranges and angles so that they have different degrees of reliability. Proper accounting for these different reliabilities has the potential to improve decision making compared to simple uniform treatment of the sensors. Consistency problems arising in majority voting are addressed with a view to achieving high classification performance. This is done by introducing preference ordering among the possible target types and assigning reliability measures (which essentially serve as weights) to each decision-making node based on the target range and azimuth estimates it makes and the belief values it assigns to possible target types. The results bring substantial improvement over evidential reasoning and simple majority voting by reducing the target misclassification rate.