基于LMKL和OC-ELM的航空电子部件故障检测方法

针对航空电子部件故障样本获取困难以及检测准确率不高的问题,提出基于局部多核学习(localized multiple kernel learning, LMKL)和一类超限学习机(one-class extreme learning machine, OC-ELM)的故障检测方法。仅运用正常状态的小样本数据,给出了LMK-OC-ELM的数学表达形式,并在不同的门模型下推导了LMK-OC-ELM中局部核权重的优化方法;在获取局部核权重的基础上,定义了离线故障检测所需的统计检验量与阈值,以便工程实现。将所提方法应用于某型接收机,结果表明,在训练时间可控的前提下,与4种常见的一类分类(one-class classification, OCC)算法相比,所提方法可均衡地提高召回率、查准率和特异度,以LMK-OC-ELM-sig为代表,其在F1、曲线下方面积(area under curve, AUC)、G-mean和准确率4个指标上,比最近提出的局部多核异常检测(localized multiple kernel anomaly detection, LMKAD)方法分别提高了1.60%、1.57%、1.53%和2.23%。     

Incremental support vector machine algorithm based on multi-kernel learning

A new incremental support vector machine (SVM) algorithm is proposed which is based on multiple kernel learning.Through introducing multiple kernel learning into the SVM incremental learning,large scale data set learning problem can be solved effectively.Furthermore,different punishments are adopted in allusion to the training subset and the acquired support vectors,which may help to improve the performance of SVM.Simulation results indicate that the proposed algorithm can not only solve the model selection problem in SVM incremental learning,but also improve the classification or prediction precision.     

联机核模糊C均值聚类方法

基于核模糊C均值(kernel fuzzy C-means, KFCM)提出了一种针对较大规模数据的联机核模糊C均值 (online kernel fuzzy C-means, OKFCM) 算法,同时考虑到核参数的选择困境,借鉴多核学习思想,进一步衍生出了联机多核模糊C均值 (online multiple kernel fuzzy C-means, OMKFCM) 算法。由此,在有效缓和核参数选择难题的同时,新算法不仅继承了KFCM优越的聚类特性且适合聚类数据流。最后,在人工和真实数据集上验证了新提出的核联机算法比现有基于划分的大规模数据处理算法具有更好的性能。     

A self-similar local neuro-fuzzy model for short-term demand forecasting

This paper proposes a selfsimilar local neurofuzzy （SSLNF） model with mutual informati onbased input selection algorithm for the shortterm electricity demand forecasting. The proposed self similar model is composed of a number of local models, each being a local linear neurofuzzy （LLNF） model, and their associated validity functions and can be interpreted itself as an LLNF model. The proposed model is trained by a nested local liner model tree （NLOLIMOT） learning algorithm which partitions the input space into axisorthogonal subdomains and then fits an LLNF model and its associated validity function on each subdomain. Furthermore, the proposed approach allows different input spaces for rule premises （validity functions） and consequents （local models）. This appealing property is employed to assign the candidate input variables （i.e., previous load and temperature） which influence shortterm electricity demand in linear and nonlinear ways to local models and validity functions, respectively. Numerical results from shortterm load forecasting in the New England in 2002 demonstrated the accuracy of the SSLNF model for the STLF applications.     

Using junction trees for structural learning of Bayesian networks

The learning Bayesian network (BN) structure from data is an NP-hard problem and still one of the most exciting challenges in the machine learning.In this work,a novel algorithm is presented which combines ideas from local learning,constraintbased,and search-and-score techniques in a principled and effective way.It first reconstructs the junction tree of a BN and then performs a K2-scoring greedy search to orientate the local edges in the cliques of junction tree.Theoretical and experimental results show the proposed algorithm is capable of handling networks with a large number of variables.Its comparison with the well-known K2 algorithm is also presented.     

基于改进稀疏KELM的在线非平稳动态系统状态预测方法

本文针对基于核的增量超限学习机(kernel based incremental extreme learning machine,KB-IELM)对非平稳动态系统的时变状态跟踪能力不足的问题,提出一种新型的状态预测方法。通过融合遗忘因子和自适应时变正则化因子构建新的目标函数。通过最小化字典的快速留一交叉验证(fast leave-one-out cross-validation, FLOO-CV)误差,选择具有预定规模的关键节点以构成字典。通过融合遗忘因子,为字典中各关键节点按时间顺序分配不同权重。基于FLOO-CV原则,使用天牛须搜索(beetle antennae search,BAS)算法为不同的非线性区域赋予不同的正则化参数。通过矩阵初等变换和分块求逆,实现核权重向量的在线递推更新。将模型应用于非平稳Mackey-Glass混沌时间序列预测和某型飞机发动机的状态预测。所提算法相比于最新的非平稳在线序列核超限学习机(non-stationary online sequential kernel extreme learning machine,NOS-KELM)和融合自适应正则化因子的在线稀疏核超限学习机(online sparse kernel extreme learning machine with adaptive regulation factor, OSKELM-ARF)两种方法,其训练精度分别提升了66.67%、50.72%、预测精度提升了67.02%、56.34%,最大预测误差减少了67.27%、51.09%,平均相对误差率分别减少了67.18%、59.62%。实验证明所提算法有效提升了在线预测的精度。     

Novel ensemble learning based on multiple section distribution in distributed environment

Because most ensemble learning algorithms use the centralized model, and the training instances must be centralized on a single station, it is difficult to centralize the training data on a station. A distributed ensemble learning algorithm is proposed which has two kinds of weight genes of instances that denote the global distribution and the local distribution. Instead of the repeated sampling method in the standard ensemble learning, non-balance sampling from each station is used to train the base classifier set of each station. The concept of the effective nearby region for local integration classifier is proposed, and is used for the dynamic integration method of multiple classifiers in distributed environment. The experiments show that the ensemble learning algorithm in distributed environment proposed could reduce the time of training the base classifiers effectively, and ensure the classify performance is as same as the centralized learning method.     

基于改进稀疏KELM的在线非平稳动态系统状态预测方法

本文针对基于核的增量超限学习机(kernel based incremental extreme learning machine，KB-IELM)对非平稳动态系统的时变状态跟踪能力不足的问题，提出一种新型的状态预测方法。通过融合遗忘因子和自适应时变正则化因子构建新的目标函数。通过最小化字典的快速留一交叉验证(fast leave-one-out cross-validation, FLOO-CV)误差，选择具有预定规模的关键节点以构成字典。通过融合遗忘因子，为字典中各关键节点按时间顺序分配不同权重。基于FLOO-CV原则，使用天牛须搜索(beetle antennae search，BAS)算法为不同的非线性区域赋予不同的正则化参数。通过矩阵初等变换和分块求逆，实现核权重向量的在线递推更新。将模型应用于非平稳Mackey-Glass混沌时间序列预测和某型飞机发动机的状态预测。所提算法相比于最新的非平稳在线序列核超限学习机(non-stationary online sequential kernel extreme learning machine，NOS-KELM)和融合自适应正则化因子的在线稀疏核超限学习机(online sparse kernel extreme learning machine with adaptive regulation factor, OSKELM-ARF)两种方法，其训练精度分别提升了66.67%、50.72%、预测精度提升了67.02%、56.34%，最大预测误差减少了67.27%、51.09%，平均相对误差率分别减少了67.18%、59.62%。实验证明所提算法有效提升了在线预测的精度。     

ON APPROXIMATION BY REPRODUCING KERNEL SPACES IN WEIGHTED L^p SPACES

In this paper, we investigate the order of approximation by reproducing kernel spaces on （-1, 1） in weighted L^p spaces. We first restate the translation network from the view of reproducing kernel spaces and then construct a sequence of approximating operators with the help of Jacobi orthogonal polynomials, with which we establish a kind of Jackson inequality to describe the error estimate. Finally, The results are used to discuss an approximation problem arising from learning theory.     

集成粗糙集和阴影集的簇特征加权模糊聚类算法

特征加权是聚类算法中的常用方法,决定权值对产生一个有效划分非常关键。基于模糊集、粗糙集和阴影集的粒计算框架,本文提出计算不同簇特征权重的聚类新方法,特征权值随着每次迭代自动地计算。每个簇采用不同的特征权重可以更有效地实现聚类目标,并使用聚类有效性指标包括戴维斯-Bouldin指标(Davies-Bouldin,DB)、邓恩指标(Dunn, Dunn)和Xie-Beni指标(Xie-Beni, XB)分析基于划分的聚类有效性。真实数据集上的实验表明这些算法总是收敛的,而且对交叠的簇划分更有效,同时在噪声和异常数据存在时具有鲁棒性。     

核模糊C均值聚类算法优选BDS-3三频组合观测值

目前对全球导航卫星系统(global navigation satellite system, GNSS)三频组合观测值优选的研究,主要集中在全球定位系统(global positioning system, GPS)和北斗二号(beidou navigation satellite system, BDS-2)上,对BDS-3的研究相对较少。为克服以往聚类优选算法中存在的仅适用于类球形簇、聚类数目和初始聚类中心的确定主观性强、对离群点敏感、易陷于局部最优等不足,提出一种改进的核模糊C均值聚类算法,引入核函数与抑制离群点的新距离度量,基于多类广义核极化准则优化核参数,用改进爬山法确定聚类数目与初始聚类中心。然后,以模糊C均值聚类算法为对照进行了对比实验,在短、长两种基线下分别解算组合模糊度。通过对优选所得代表性组合的模糊度固定成功率进行对比分析,验证了该算法的可行性与算法改进的有效性。     

基于PS-IFKCM的弹道中段目标识别方法

针对现有直觉模糊核c-均值(intuitionistic fuzzy kernel c-means,IFKCM)聚类算法对初始值敏感、易陷入局部最优解及收敛速度慢等缺陷,汲取了粒子群优化(particle swarm optimization,PSO)算法优势,对初始聚类中心进行优化,提出了基于粒子群优化的直觉核c-均值(particle swarm-based intuitionistic fuzzy kernel c-means,PS-IFKCM)聚类算法,选取4组标准数据集实际样本数据对算法的有效性进行了试验。最后选取弹道中段目标识别常用的雷达截面积(radar cross section, RCS)这一特征属性进行弹道中段目标识别仿真实验,并将其与模糊c-均值（fuzzy c-means, FCM）算法、IFKCM算法的识别效果及运行时间进行比较分析,表明了该算法应用于弹道中段目标识别的有效性及优越性。     

改进的核子类判决分析

提出了改进的核子类判决分析（improved kernel clustering based discriminant analysis, IKCDA）方法,首先采用快速全局核k 均值聚类算法找到每类目标的最优子类划分,然后基于找到的子类划分结果采用核子类判决分析求取最优的投影矢量。基于UCI机器学习数据库的实验结果表明,经过IKCDA特征提取后异类样本间的可分性明显改善了。此外,基于美国运动和静止目标获取与识别（moving and stationary target acquisition and recognition, MSTAR）计划录取的合成孔径雷达地面静止目标数据的实验结果表明,经过IKCDA后可以改善对真实目标的分类性能和对干扰目标的拒判能力。     

航电系统体系贡献率权重演化动态综合评估

随着应用需求的发展,航电系统建设的体系特征日益明显,开展航电系统体系贡献率评估成为引导其迭代更新与优化设计的关键。针对专家知识随时间积累以及蜂群、协同等作战场景变化带来的指标体系权重演化问题,提出了一种适用于航电系统体系贡献率多阶段评估的动态综合方法。基于航电系统任务能力要素,构建了体系贡献率评估指标体系,并应用粒子群优化算法实现了有效评估阶段的时间加权。与传统静态单次评估、基于熵权法(entropy weight method, EWM)的动态评估以及逼近理想解排序法(technique for order preference by similarity to an ideal solution, TOPSIS)的动态评比等对比,所提方法充分体现权重分配信息并统筹兼顾阶段时序差异,能够更准确地反映指标贡献权重和能力贡献分布等评估结果,从而为航电系统发展论证提供更可靠和更灵活的决策方法支持。     

考虑全过程优化的支持向量机预测方法

针对支持向量机（support vector machine, SVM）预测过程中影响因素选择、输入特征集优化、核函数选择及参数优化方面存在的问题,提出了一种全过程优化方法。首先使用频繁模式增长关联规则分析和模糊贝叶斯网络组合模型来解决影响因素选择中存在的主观性问题,然后使用在异常值处理和类内距离与类间距离方面进行改进的模糊C均值聚类算法优化输入特征集,减小支持向量机预测模型冗余度及训练样本集过修正度,通过比较各核函数的特点选择径向基核函数作为SVC的核函数,改进了粒子群优化算法中微粒速度和位置函数及惯性权重值算法,使用该方法优化SVM参数并建立预测模型。最后，通过案例运算和分析,证明该文方法具有更高的预测精度。     

锌电解过程电流效率的模糊神经网络模型设计

针对锌电解过程参数关系的非线性,样本数据少,知识有限等特点,本文采用了一种五层结构的模糊神经网络建立电流效率与酸锌比、电流密度的关系模型;为避免神经网络学习过程陷入局部最小,首先聚类产生网络的初始值;然后通过一种基于模糊逻辑的启发式学习算法对神经网络进行训练,提高收敛速度;仿真结果表明了这种模型的有效性.     

自适应特征熵权模糊C均值聚类算法的研究

特征权重算法对聚类效果有很大的影响,而传统的特征权重算法忽略了特征项在类间和类内的分布情况.因此,研究聚类后样本特征属性表现的有序性程度对聚类结果的影响,分析聚类后样本特征属性的分布情况,提出了一种自适应特征熵权模糊C均值聚类算法.该算法以聚类后的特征熵和信息增益作为准则调整特征权值,通过聚类与权重更新逐步迭代优化,直至获得最优的特征权值.实验表明,自适应特征熵权模糊C均值聚类算法能够有效地区分各个特征属性对聚类效果的重要程度;较于其它加权模糊C均值聚类算法,该算法能够得到更高的聚类准确率.