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.     

带有工况中心修正的多模型在线建模

针对运行工况频繁波动、单一模型难以描述过程特性的问题,提出了带有工况中心修正的多模型在线建模方案,包括工况识别机制、局部模型、多模型合成机制.工况识别机制根据工况特征变量分析工况范围,由相近度修正工况中心;局部模型采用Hammerstein模型,非线性增益由带有稳定学习算法的小波神经网络建立,线性模型由带控制量的自回归模型(ARX)建立;多模型合成机制采用加权求和方法.在线修正工况中心可反映工况的时间变化特性,参数稳定学习算法改善了模型精度和自适应能力.采用此方法建立污水处理过程化学需氧量(COD)软测量模型,结果表明,模型在工况大范围变化时仍具有满意预测效果.     

基于自适应模糊聚类多模型过程监控及应用

对乙烯裂解炉建立实时监控模型具有重要的现实意义,而传统的多元统计过程监控方法都是假设过程处于单一工况下,而随着过程参数(进料负荷、产品组分等)的改变,工况也随之改变,传统方法便不再适用.本文针对工业过程中的多工况问题,提出了一种基于自适应模糊聚类的多模型过程监控方法,该方法可以减少监控方法对过程知识的依赖性,并且能够适应实际工业过程的非高斯性和非线性特征.首先对影响工况的过程变量利用自适应模糊聚类进行工况划分,然后对每种工况的建模数据分别利用最大方差展开(MVU)提取低维信息,再用支持向量数据描述(SVDD)建立多模型过程监控模型,最后再利用相应的统计指标进行过程监控.将上述方法应用在乙烯裂解炉上,并与基于高斯混合模型的多PCA方法(GMM-MPCA)进行了比较.仿真实验中,监控对裂解炉运行影响最大的33个变量,根据聚类有效性指标,将数据划分为5类时可以得到最佳的聚类效果.通过实验,将33维建模数据降到20维时误报率最小.仿真结果表明该方法在对非线性和非高斯性过程的监控上,能达到很好的效果,误报率和检测率均优于GMM-MPCA方法.     

基于GMM的多工况过程监测方法

传统基于主元分析的故障检测方法大多假设工业过程只运行在1个稳定工况,数据服从单一的高斯分布。若这些方法直接用于多工况过程则将会产生大量的误检。为此,本文提出了1种基于高斯混合模型的多工况过程监测方法。首先利用PCA变换对过程数据集进行降维,在主元空间建立高斯混合模型对过程数据进行聚类,自动获取工况数和相关分布特性。然后对每个工况建立主元分析(principal component analysis,PCA)模型来描述整个运行过程数据分布的统计特性。最后在过程监测中,根据监测样本属于各个工况的概率构造综合统计量,实现对多工况过程的故障检测。TE过程的仿真结果表明,本文提出的方法与传统的PCA方法相比,能自动获取工况和精确估计各个工况的统计特性,从而能更准确及时地检测出多工况过程的各种故障。     

基于特征空间k最近邻的批次过程监视

针对具有非高斯、非线性及多工况特性的批次过程,提出一种基于特征量最近邻统计指标的过程监视方法. 首先,将批次过程正常工况原始数据投影到其特征空间,提取主元T和平方预测误差SPE,并进行特征量k最近邻距离平方和的求解. 然后,采用核密度估计法获得概率密度分布函数,确定统计监视控制限. 特征空间的主元T和SPE特征量能全面代表原始数据的有用信息. 采用特征量k最近邻建立监视模型将会节省存储空间,提高建模样本数量与变量之比以及检测异常工况的速度. 另外,利用局部近邻数据建模可以解决过程具有的非线性和多工况问题,而应用核密度估计法可以解决过程数据具有的非高斯分布问题. 最后,在半导体生产过程的成功应用表明了所提方法的有效性.     

基于GLRGMM的间歇过程在线监控策略

针对间歇过程的非线性和动态性,提出了全局—局部正则化高斯混合模型 （GLRGMM）算法。首先引入邻域保持嵌入算法提取局部流形结构,通过寻求一种低维投影对非线性过程进行全局结构保持,同时最大限度地保留局部流形特征;然后通过对高斯混合模型引入正则项来在线监控更新高斯模型,获取非线性数据流形结构,解决数据动态性问题;最后集成全局—局部监控指标实现在线监控。通过青霉素发酵过程进行了验证,结果表明所提算法比DPCA、GLNPE具有更好的在线监控效果。     

独立成分相关分析的自适应故障监测方法

工业过程数据具有动态、非高斯等特性.独立成分分析(independent component analysis, ICA)既可以分析数据的非高斯形式,又可以极大地去除多变量间的耦合且满足独立性要求.本文引入粒子群算法优化ICA模型参数,自适应地确定独立成分个数.同时,提出一种基于隐马尔科夫链模型(hidden Markov model, HMM)的自适应检测限设计方法,将时间相关数据块的特征信息变化作为过程故障的检测依据.首先利用由时间窗方法确定的独立成分组成监测矩阵来训练HMM模型,旨在提高独立成分间相关性水平的表示能力;然后将得到的HMM模型对监测矩阵进行相关性评估,并在一定容许裕度的基础上设计评估值的自适应因子及检测限,并据此监测特征信息变化,动态地进行在线故障检测.最后, Tennessee Eastman (TE)仿真平台的实验结果表明了所提方法的有效性.     

基于减法聚类的多模型在线辨识算法

考虑到实际工业过程中复杂系统的工况变化往往具有不确定性的特点, 离线辨识的多模型系统难以自适应反映系统的非线性, 因此本文提出一种新的基于减法聚类的多模型在线辨识算法. 首先采用在线聚类算法辨识多模型系统中的局部模型个数与工况参数, 然后充分考虑聚类发生变化对局部模型参数辨识的影响, 给出相应的局部模型参数在线辨识算法. 最后以某电厂300MW锅炉--汽轮机的协调控制系统为对象, 采用上述辨识方法进行仿真研究, 结果验证了本文算法的有效性.     

基于修正闭环子空间辨识–分段线性结构的环管式丙烯聚合反应过程非线性模型预测控制（英文）

针对环管式聚丙烯生产过程装置多变量、耦合和非线性等特性容易导致过程控制不稳定及质量指标波动问题,本文提出了一种基于修正闭环子空间辨识–分段线性(MSSARX--PWL)维纳(Wiener)模型结构的非线性模型预测控制算法.利用修正的闭环子空间辨识方法(MSSARX)辨识对象在闭环工况下的线性状态空间模型,并将该线性模型与多变量分段线性化(PWL)方法辨识得到的非线性稳态模型结合,建立双环管丙烯聚合反应动态过程的非线性预测模型,而后进一步将非线性模型转化为线性模型,在线性预测控制算法框架下用二次线性规划方法(LQP)优化控制器,无须用非线性规划方法(NLP)求解.从双环管丙烯聚合反应过程仿真例子表明,该算法不仅能保证模型和控制精度,而且能提高计算效率.     

基于四阶累积分析的工业大肠杆菌制备过程故障诊断

工业大肠杆菌制备过程具有非线性和非高斯性共存的特征,导致难以对故障源进行有效定位,针对这个问题,提出一种基于多向核熵独立元分析(MKEICA)的过程监测方法;同时针对传统低阶监控统计量(T2, I2和SPE)无法得到非高斯信息的不足提出了四阶累积监控统计量的方法;其次通过对四阶累积监控量进行推导,得到故障产生的原因.最后将其应用在实际的工业过程并与多向核独立元分析(MKICA)监测模型进行对比验证该方法的可行性及有效性.     

Quality relevant nonlinear batch process performance monitoring using a kernel based multiway non-Gaussian latent subspace projection approach

Multiway kernel partial least squares method (MKPLS) has recently been developed for monitoring the operational performance of nonlinear batch or semi-batch processes. It has strong capability to handle batch trajectories and nonlinear process dynamics, which cannot be effectively dealt with by traditional multiway partial least squares (MPLS) technique. However, MKPLS method may not be effective in capturing significant non-Gaussian features of batch processes because only the second-order statistics instead of higher-order statistics are taken into account in the underlying model. On the other hand, multiway kernel independent component analysis (MKICA) has been proposed for nonlinear batch process monitoring and fault detection. Different from MKPLS, MKICA can extract not only nonlinear but also non-Gaussian features through maximizing the higher-order statistic of negentropy instead of second-order statistic of covariance within the high-dimensional kernel space. Nevertheless, MKICA based process monitoring approaches may not be well suited in many batch processes because only process measurement variables are utilized while quality variables are not considered in the multivariate models. In this paper, a novel multiway kernel based quality relevant non-Gaussian latent subspace projection (MKQNGLSP) approach is proposed in order to monitor the operational performance of batch processes with nonlinear and non-Gaussian dynamics by combining measurement and quality variables. First, both process measurement and quality variables are projected onto high-dimensional nonlinear kernel feature spaces, respectively. Then, the multidimensional latent directions within kernel feature subspaces corresponding to measurement and quality variables are concurrently searched for so that the maximized mutual information between the measurement and quality spaces is obtained. The I² and SPE monitoring indices within the extracted latent subspaces are further defined to capture batch process faults resulting in abnormal product quality. The proposed MKQNGLSP method is applied to a fed-batch penicillin fermentation process and the operational performance monitoring results demonstrate the superiority of the developed method as apposed to the MKPLS based process monitoring approach.     

Process monitoring through manifold regularization-based GMM with global/local information

The nonlinear and multimodal characteristics in many manufacturing processes have posed some difficulties to regular multivariate statistical process control (MSPC) (e.g., principal component analysis (PCA)-based monitoring method) because a fundamental assumption is that the process data follow unimodal and Gaussian distribution. To explicitly address these important data distribution characteristics in some complicated processes, a novel manifold learning algorithm, joint local intrinsic and global/local variance preserving projection (JLGLPP) is proposed for information extraction from process data. Based on the features extracted by JLGLPP, local/nonlocal manifold regularization-based Gaussian mixture model (LNGMM) is proposed to estimate process data distributions with nonlinear and multimodal characteristics. A probabilistic indicator for quantifying process states is further developed, which effectively combines local and global information extracted from a baseline GMM. Thus, the JLGLPP and LNGMM-based monitoring model can be used effectively for online process monitoring under complicated working conditions. The experimental results illustrate that the proposed method effectively captures meaningful information hidden in the process signals and shows superior process monitoring performance compared to regular monitoring methods.     

基于互信息的多块k近邻故障监测及诊断

由于传统的k近邻故障监测不考虑过程的局部信息,只建立一个全局模型,因此提出一种基于互信息的多块k近邻故障监测方法。首先,考虑建模数据的非线性和非高斯等特性,基于变量间的互信息进行子块构建;然后,利用k近邻方法对每个子块进行建模与监测,子块中的k近邻模型反映了更多的过程局部特征;最后,将所有子块的监测结果通过贝叶斯推断方法进行融合,并采用基于马氏距离的故障诊断方法辨识故障源。通过对田纳西-伊斯曼过程和高炉炼铁过程中的应用仿真,监测结果表明所提方法的可行性和有效性。     

基于GMM的多模态过程模态识别与过程监测

多模态复杂过程的多变量、多工序、变量时变性以及模态转换时间不确定等多种原因, 导致面向多模态生产过程的监测问题十分复杂. 对此, 基于高斯混合模型的监测方法, 结合定性知识和定量知识, 解决了多模态过程监测中离线数据模态划分、稳定模态和过渡模态的监测模型建立以及在线数据的模态识别等关键问题, 最终实现了对多模态过程的监测.     

在线鲁棒最小二乘支持向量机回归建模

鉴于工业过程的时变特性以及现场采集的数据通常具有非线性特性且包含离群点,利用最小二乘支持向量机回归（least squares support vector regression,LSSVR）建模易受离群点的影响.针对这一问题,结合鲁棒学习算法（robust learning algorithm,RLA）,本文提出了一种在线鲁棒最小二乘支持向量机回归建模方法.该方法首先利用LSSVR模型对过程输出进行预测,与真实输出相比较得到预测误差;然后利用RLA方法训练LSSVR模型的权值,建立鲁棒LSSVR模型;最后应用增量学习方法在线更新鲁棒LSSVR模型,从而得到在线鲁棒LSSVR模型.仿真研究验证了所提方法的有效性.     

基于k近邻的批次过程在线实时监测方法

针对基于k近邻的故障检测方法(Fault Detection method using the k-Nearest Neighbor rule,FD-kNN)的在线实时监测需预估当前时刻之后的采样数据,检测性能会受到预估精度影响的问题,对FD-kNN进行扩展以适用于批次过程的实时监测.该方法根据每个采样时刻的历史数据进行建模,并根据这些模型实时监测批次过程.该方法不需要预估数据,避免由于预估误差大而带来的误报和漏报问题,同时较好地继承k近邻法则(k-Nearest Neighbor rule,kNN)在处理非线性、多模态和非高斯等问题上具有的优势.青霉素发酵过程的仿真试验验证该方法可行.     

Multivariate statistical monitoring of two-dimensional dynamic batch processes utilizing non-Gaussian information

Dynamics are inherent characteristics of batch processes, and they may exist not only within a particular batch, but also from batch to batch. To model and monitor such two-dimensional (2D) batch dynamics, two-dimensional dynamic principal component analysis (2D-DPCA) has been developed. However, the original 2D-DPCA calculates the monitoring control limits based on the multivariate Gaussian distribution assumption which may be invalid because of the existence of 2D dynamics. Moreover, the multiphase features of many batch processes may lead to more significant non-Gaussianity. In this paper, Gaussian mixture model (GMM) is integrated with 2D-DPCA to address the non-Gaussian issue in 2D dynamic batch process monitoring. Joint probability density functions (pdf) are estimated to summarize the information contained in 2D-DPCA subspaces. Consequently, for online monitoring, control limits can be calculated based on the joint pdf. A two-phase fed-batch fermentation process for penicillin production is used to verify the effectiveness of the proposed method.     

Phase analysis and statistical modeling with limited batches for multimode and multiphase process monitoring

For multimode batch processes, the conventional modeling methods in general require that sufficient batches should be available for every mode, which, however, cannot be guaranteed in practice. It may be impractical to conduct enough trial runs and wait until sufficient batches are available before development of monitoring models for each mode. Starting from limited batches, how to derive reliable process information and develop monitoring models has been an important question for successful online multimode batch process monitoring. To address this problem, this article proposes a phase analysis and statistical modeling strategy with limited batches. One mode which has obtained sufficient batches is chosen as the reference mode while the other modes which can only get limited batches work as alternative modes. Starting from limited batches, the proposed algorithm addresses two issues, concurrent phase partition and analysis of between-mode relative changes. First, for each alternative mode, generalized time-slices are constructed by combining several consecutive time-slices within a short time region to explore local process correlations. The time-varying characteristics are then concurrently analyzed across modes so that multiple sequential phases are identified simultaneously for all modes. Then phase-representative data units are arranged by variable-unfolding the conventional time-slices for the reference mode and the generalized time-slices for each alternative mode respectively. Between-mode statistical analysis is performed within each phase where the relative changes from the reference mode to each alternative mode are analyzed. From the between-mode perspective, different types of relative variations in each alternative mode are separated and modeled for online monitoring. Starting from limited batches, online batch process monitoring can be conducted, providing reliable fault detection performance. The proposed algorithm is illustrated with a typical multiphase batch process with multiple modes.