基于混合PCA模型的多工况过程监控方法

传统的多变量统计过程监控方法一般都假设过程只运行在一个稳定工况下,但很多实际工业过程往往具有多工况特征.针对这一问题,提出一种基于混合PCA模型的多工况过程监控方法.将混合高斯模型和PCA相结合,用改进的EM算法估计模型的工况数以及各工况的分布参数和主元数,并构建归一化的统计量实现对多工况过程的监控.TE过程的仿真研究表明,所提出的方法相对传统PCA方法能更精确地估计各工况的统计特性,从而更准确及时地检测出多工况过程的各种故障.     

一种基于GMM的多工况过程故障诊断方法

故障发生后,为了查出故障原因,需要辨识出故障变量实现故障诊断。传统多元统计故障诊断方法通常假设过程只运行在单一工况下,从而不能适用于多工况过程。针对这个问题,提出了1种基于高斯混合模型(Gaussian Mixture Models,GMM)的多工况化工过程故障诊断方法。首先建立GMM对过程数据进行聚类,自动估计出工况数以及各工况的分布参数,同时对数据进行标准化处理,使得各个变量在后续处理中占有相同比重,然后对各个工况建立PCA(PrincipalComponentAnalysis)模型。当检测到故障发生后,本文通过结合相对贡献量和后验概率,构造全局相对贡献量得到全局相对贡献图,用于多工况过程故障变量的识别。CSTH过程的仿真结果表明,本文提出的方法能够有效地辨识出故障变量。     

基于高斯分量标准化的K近邻故障检测策略

针对复杂多工况工业过程故障检测问题,提出一种基于高斯分量标准化的K近邻(Gaussian Component Standardization K-Nearest Neighbor, GCS-KNN)故障检测策略。样本数据应用高斯混合模型(Gaussian Mixture Model, GMM)进行训练,将数据分解为多个高斯分量;通过每个高斯分量的均值和协方差对该分量内的数据进行标准化处理;应用K近邻(K-Nearest Neighbor, KNN)算法对标准化后的样本进行检测。GCS-KNN通过高斯分量标准化消除数据的多模态特性,提高传统基于KNN检测方法的检测率。利用数值例子和半导体工业过程仿真实验验证了该方法的有效性,并与传统的主元分析(Principal Component Analysis, PCA)、KNN、动态主元分析(Dynamic PCA,DPCA)和加权KNN(Weighted KNN,WKNN)等方法进行对比,结果证实此方法具有显著的优势。     

基于Q统计量分量的故障检测算法研究

针对传统PCA故障检测算法的结果有定论不明确的缺陷,提出一种基于Q统计量分离的故障检测新方法,把Q统计量分为PVR和CVR,前者代表显著与主元有关的变量信息,后者代表与主元无明显关系的变量信息,再配合T2统计量共同用于监测过程,检测效果更细致.将此方法结合基于累积方差贡献率(CPV)和复相关系数(MCC)确定过程监测模型主元数的新方法,监测β-甘露聚糖酶发酵工业的过程,与传统的PCA故障检测方法比较,仿真研究结果表明该算法能够确保主元空间(PCS)中的信息存量,充分刻画过程变化,有效识别正常工况变化与故障,正确检测微弱故障,提高过程监控的准确性.     

一种基于多工况识别的过程在线监测方法

在包含多个工况的工业生产过程中,各个稳态工况之间存在着一定的过渡过程,虽然过渡时间较短,但其复杂的动态特性使得传统的过程监测方法难以获得满意的效果,为此提出一种基于多工况识别的过程监测方法.首先,通过窗口切割对基本稳态工况进行识别;然后,采用滑动窗技术确定过渡过程的起始和结束时间,并进一步基于差分分段技术对过渡过程的子阶段进行分类,考虑到各阶段数据的不同分布特性,利用独立成分分析和主元分析分别提取各阶段数据的非高斯和高斯信息;最后根据贝叶斯推断将3个统计量进行重构, 实现多工况过程的在线监测.通过TE过程的仿真研究,验证了所提出方法的可行性和有效性.     

基于动态主元分析的空分过程异常工况在线诊断

空分过程异常工况的在线诊断对于保证空分产品质量、降低"氮塞"等故障的发生率和减小故障损失有着重要的实用价值。主元分析(PCA)是1种数据驱动的统计建模方法,已广泛应用于复杂工业过程的运行状态监控和故障诊断,然而,传统的PCA方法不能够反应数据的动态特性。动态主元分析(DPCA)作为1种将传统PCA推广到多变量动态过程的方法,其时滞长度的确定是DPCA的关键点。本文应用动态主元分析(DPCA)方法建立了空分过程异常工况的在线诊断系统,并结合空分过程的故障诊断特性,对DPCA中时滞长度提出了可行的确定方法。实际运行效果表明该系统对故障的报警率为100%,误报率约4%,证明了文中所建立的诊断系统的有效性及文中所提出的对于复杂连续生产系统确定DPCA时滞长度的方法的有效性及可行性。     

基于缺失数据的误差生成策略及其在故障检测中的应用

误差生成是基于机理模型故障检测方法的核心本质,但鲜有应用于统计过程监测方法中.为此,提出一种基于缺失数据的误差生成策略,将能反映出采样数据对统计模型拟合程度的误差作为新的被监测对象实施故障检测.所提出的基于缺失数据的主元分析(MD-PCA)方法通过逐一假设各变量测量数据缺失后,利用缺失数据处理方法推测出相应缺失数据的估计值,并对缺失数据的实际值与估计值之间的误差实施基于PCA模型的故障检测.利用误差实施故障检测的优势在于,生成的误差能在一定程度上降低原测量变量的非高斯性程度,而且误差体现的是对应缺失变量中与其他测量变量不相关的成分信息,更能揭示各测量变量的本质.通过在TE过程上的实验充分验证了所提出方法的优势,以及MD-PCA方法用于故障检测的可行性与优越性.     

基于PC-wkNN的多工况间歇过程故障检测方法研究*

针对多工况间歇生产过程中,过程数据维数高、中心漂移和方差差异明显等特点,提出了基于主元分析和加权k近邻相结合的故障检测方法(PC-wkNN)。首先应用PCA确定训练数据主元模型,简约数据结构;其次在主元空间以训练样本的第k近邻的局部近邻平均距离倒数为权重,构建加权距离D.加权距离D可以有效降低工况中心漂移和方差差异明显的影响。最后,根据t分布确定统计量D的控制限,当测试样本加权距离D大于控制限,则其为故障;否则为正常。PC-wkNN提高了工况间歇过程故障检测率。通过两个模拟实例及青霉素发酵仿真实验,与PCA,KPCA,FD-kNN,PC-kNN等方法比较,验证了本文方法的有效性。     

化工过程故障原因诊断的变量异常顺序法

为对化工过程故障进行实时诊断,建立设了备的主元分析(Principal component analysis,PCA)模型,根据实时数据和PCA模型计算综合指标以在线检测其故障的发生,并提出了PCA模型的在线更新策略,以适应实际过程中工况缓变特性。为在线监测到故障发生时,能确定故障根原因,根据各变量的DCS报警上下限判断其异常状态,并记录各变量出现异常的时间顺序,以供操作人员参考,从而准确地诊断所发生故障的根源。基于过程安全生产指导平台,将所提出的方法实际应用于某炼油厂延迟焦化装置的分馏塔单元,长期在线应用结果表明所提出的在线更新PCA模型能准确地连续检测出故障的发生并适应工况的缓变,而变量异常顺序可帮助操作人员正确地确定故障原因。     

基于多尺度核主元分析的非线性过程监控方法研究

针对化工过程数据的多尺度性和非线性特性,提出了一种多尺度核主元分析方法(MSKPCA)监控过程的运行状态。使用小波变换在不同尺度下分解测量信号．然后借助于核函数对分解后的数据进行非线性变换,在变换后的线性空间中用主元分析(PCA)提取过程数据的主要特征,构造监控统计量T2和Q来检测故障。在此基础上,提出了一种贡献图方法．计算过程变量对故障的贡献量,用于故障变量的分离。在TE过程上的监控结果表明,MSKPCA可以比PCA和动态PCA更迅速地检测到过程故障,贡献图方法能够正确地分离故障变量。     

基于两层分块GMM-PRS的流程工业过程运行状态评价

过程运行状态评价旨在实时判断运行性能优劣程度,并追溯导致非优运行状态的原因,指导操作人员进行生产调整,保证企业经济效益.因此,对过程运行性能优劣评价的研究具有重要的理论和应用价值.本文针对定量、定性变量共存的流程工业过程运行状态评价问题,提出基于两层分块混合模型的评价方法.将流程工业过程根据其物理特性和管理方向划分子块,产生子块层和全流程层.在定量信息占主导地位的子块内,建立定量的高斯混合模型（Gaussian mixture model,GMM）.在定性信息占主导地位的子块内,建立定性概率粗糙集（Probabilistic rough set,PRS）模型.综合各子块运行状态信息,进一步判定全流程运行状态等级.针对非优运行状态等级,本文提出基于贡献率的非优原因追溯方法,在非优子块内进行原因追溯.最后,将所提方法应用于某黄金湿法冶炼生产过程,说明所提方法的可行性和有效性.     

Multimode process monitoring with PCA mixture model

For multimode processes, Gaussian mixture model (GMM) has been applied to estimate the probability density function of the process data under normal-operational condition in last few years. However, learning GMM with the expectation maximization (EM) algorithm from process data can be difficult or even infeasible for high-dimensional and collinear process variables. To address this issue, a novel multimode process monitoring approach based on PCA mixture model is proposed. First, the PCA technique is directly applied to the covariance matrix of each Gaussian component to reduce the dimension of process variables and to obtain nonsingular covariance matrices. Then the Bayesian Ying-Yang incremental EM algorithm is adopted to automatically optimize the number of mixture components. With the obtained PCA mixture model, a novel process monitoring scheme is derived for fault detection of multimode processes. Three case studies are provided to evaluate the monitoring performance of the proposed method.     

Scribble-based object segmentation with modified gaussian mixture models

In this paper, we present an interactive segmentation method, designed to help the user to extract an object of interest from an image. The proposed approach adopts the scribble-based segmentation paradigm. The user interaction consists of specifying a set of lines, corresponding to both foreground and background scribbles. The segmentation process is based on color distributions, estimated with Gaussian mixture models (GMM). We show that such a technique presents some limitations when dealing with compressed images, even for relatively high quality compression factors: in this case, blocking artifacts may degrade the segmentation results. In order to overcome such a drawback, a modified GMM model, which re-shapes the Gaussian mixture based on the eigenvalues of the GMM components, is proposed. The experimental evaluation, carried out on a corpus of various images with different characteristics and textures, demonstrates the superiority of the modified GMM model which is able to appropriately take into account compression artifacts.     

基于微波雷达回波信号的智能车道划分方法

利用多目标交通测速雷达进行交通执法时,只有正确地判断出车辆所在的车道,抓拍照片才能作为交通执法的依据。传统的分车道方法主要通过人工测量的固定阈值以及坐标系旋转的方法来达到车道划分的目的,但这种方法误差较大并且不易于操作。基于统计和密度特征的核聚类算法（K-CSDF）分两步进行：首先对雷达获取的车辆数据进行特征提取,包括基于统计特征的阈值处理和基于密度特征的动态半径提取;然后引入基于核的相似性的动态聚类算法对筛选出的有效点进行聚类。通过和高斯混合模型（GMM）算法以及自组织映射神经网络（SOM）算法进行仿真对比表明：当只取100个有效点进行聚类时,K-CSDF和SOM算法能达到90%以上的分车道正确率,而GMM算法不能给出车道中心线;在算法用时上,当取1000个有效点时,K-CSDF和GMM算法用时均小于1s,可以保证实时性,而SOM算法则需要2.5s左右;在算法鲁棒性上,K-CSDF对不均匀样本的适应性优于这两种算法。当取不同数量的有效点进行聚类时,K-CSDF可以达到95%以上的平均分车道正确率。     

Optimised Learning from Demonstrations for Collaborative Robots

The approach of Learning from Demonstrations (LfD) can support human operators especially those without much programming experience to control a collaborative robot (cobot) in an intuitive and convenient means. Gaussian Mixture Model and Gaussian Mixture Regression (GMM and GMR) are useful tools for implementing such a LfD approach. However, well-performed GMM/GMR require a series of demonstrations without trembling and jerky features, which are challenging to achieve in actual environments. To address this issue, this paper presents a novel optimised approach to improve Gaussian clusters then further GMM/GMR so that LfD enabled cobots can carry out a variety of complex manufacturing tasks effectively. This research has three distinguishing innovative characteristics: 1) a Gaussian noise strategy is designed to scatter demonstrations with trembling and jerky features to better support the optimisation of GMM/GMR; 2) a Simulated Annealing-Reinforcement Learning (SA-RL) based optimisation algorithm is developed to refine the number of Gaussian clusters in eliminating potential under-/over-fitting issues on GMM/GMR; 3) a B-spline based cut-in algorithm is integrated with GMR to improve the adaptability of reproduced solutions for dynamic manufacturing tasks. To verify the approach, cases studies of pick-and-place tasks with different complexities were conducted. Experimental results and comparative analyses showed that this developed approach exhibited good performances in terms of computational efficiency, solution quality and adaptability.     

基于自适应GMM的X射线图像智能检测

传统金属工业构件X射线图像检测手段主观性过强、检测效率低下。为此,提出一种基于高斯混合模型(GMM)的智能检测方法。对同一构件的图像序列进行在线学习,每一像素点由多个高斯分布分量组成。正常工作时对每一像素点用学习到的高斯分量进行模式分 类,若不符合任一现有高斯分量就视为前景目标(损伤点),采用种子生长法连通损伤区域,确定整个损伤区域。实验结果表明,该方法可精确定位构件损伤部位,实现金属构件损伤的自动检测,检测效率较高。     

Enhanced Gaussian mixture model of RSSI purification for indoor positioning

Received Signal Strength Indicator (RSSI) is used in indoor positioning for measuring object distance to the base station. However, acquiring accurate RSSI values is challenging due to wireless interferences, especially multipath decline. Therefore, RSSI purification is critical so we propose an Enhanced Gaussian Mixture Model (EGMM) for accurate indoor positioning. EGMM improves Gaussian filter based on Gaussian Mixture Model (GMM) to identify the desirable Light of Sight (LOS) signals from multipath decline signals. EGMM integrates GMM with Akaike information criterion (AIC) to determine the best K value for GMM and filter the LOS signal set accurately. The experiment is conducted with iBeacon devices, and the average error distance of EGMM can be enhanced as 64% of those generated by existing Gaussian filtering. The average positioning error of EGMM is about 0.48 m, which is adequate for indoor positioning.     

Implementation of Real-Time Vehicle Tracking in City-Scale Video Network

Tracking unexpected warning vehicles is required for quick response to security incident. To realize real-time vehicle tracking in a large-scale video surveillance network, a geospatial and temporal connection (GSTC) model is introduced to model the connection between videos. The transition time between videos is modeled by a Gaussian mixture model (GMM). With the developed plug-ins based on GSTC and GMM, the video streams of defined geospatial neighbors are automatically called in with the video stream that the object appears in during the tracking process. Experiments show that the ratio of success of real-time tracking is largely increased.     

Efficient Speaker Recognition Using Approximated Cross Entropy (ACE)

Techniques for efficient speaker recognition are presented. These techniques are based on approximating Gaussian mixture modeling (GMM) likelihood scoring using approximated cross entropy (ACE). Gaussian mixture modeling is used for representing both training and test sessions and is shown to perform speaker recognition and retrieval extremely efficiently without any notable degradation in accuracy compared to classic GMM-based recognition. In addition, a GMM compression algorithm is presented. This algorithm decreases considerably the storage needed for speaker retrieval.     

基于Gabor 变换的GrabCut 纹理图像分割

多数自然图像都包含纹理信息, 它相对颜色特征而言具有描述方向性与尺度差异的特性. 因此, 可以利用半交互式的GrabCut 的图像分割方式对图像前景区域与背景区域进行有效的分割, 通过建立前景和背景所对应的高斯混合模型(GMM), 结合最大流最小割的图像分割方式实现全局优化, 并利用前景和背景的KL 测度, 自适应地终止分割过程. 实验对比分析表明, 所提出的方法对于合成纹理图像与自然纹理图像具有较好的整体分割效果及较高的分割准确率.