In this study, a novel Bayesian robust mixture factor analyzer (BRMFA) is proposed to deal with the robust multimode process modeling problem. Traditional factor analyzers with Gaussian assumptions are susceptible to ...
详细信息
In this study, a novel Bayesian robust mixture factor analyzer (BRMFA) is proposed to deal with the robust multimode process modeling problem. Traditional factor analyzers with Gaussian assumptions are susceptible to outliers. For this issue, the Student's t mixture model is developed so that outliers can be well explained during the modeling phase. To deal with the model selection problems, two probabilistic determination stages are merged in the Bayesian robust model. Specifically, the truncated stick-breaking represented Dirichlet process mixture (DPM) model is utilized to conduct the mixture components automatic selection, and then the automatic relevance determination (ARD) strategy is included to choose the latent space dimensions. To derive a computational tractable inference, a variational Bayesian (VB) algorithm is developed for parameter estimation. Several case studies are given for demonstrations, results of which show that the new proposed method is more insensitive to outliers during processmodeling, compared with traditional methods. (C) 2015 Elsevier B.V. All rights reserved.
Nonlinear and multimode characteristics commonly appear in modern industrial process data with increasing complexity and dynamics, which have brought challenges to soft sensor modeling. To solve these issues, in this ...
详细信息
Nonlinear and multimode characteristics commonly appear in modern industrial process data with increasing complexity and dynamics, which have brought challenges to soft sensor modeling. To solve these issues, in this article, a dynamic mixture variational autoencoder regression model is first proposed to handle the multimode industrial processmodeling with dynamic features. Furthermore, to deal with the partially labeled process data with rare quality values and large-scale unlabeled samples, a semi-supervised mixture variational autoencoder regression model is proposed, where a corresponding semi-supervised data sequence division scheme is introduced to make full use of the information in both labeled and unlabeled data. Finally, to verify the feasibility and effectiveness of the proposed methods, the models are applied to a numerical case and a methanation furnace case. The results show that the proposed methods have superior soft sensing performance, compared with the state-of-the-art methods.
In real industrial processes, factors, such as the change in manufacturing strategy and production technology lead to the creation of multimode industrial processes and the continuous emergence of new modes. Although ...
详细信息
In real industrial processes, factors, such as the change in manufacturing strategy and production technology lead to the creation of multimode industrial processes and the continuous emergence of new modes. Although the industrial SCADA system has accumulated a large amount of historical data, which can be used for modeling and monitoring multimodeprocesses to a certain extent, it is difficult for the model learned from historical data to adapt to emerging modes, resulting in the model mismatch. On the other hand, updating the model with data from new modes allows the model to continuously match the new modes, but it may cause the model to lose the ability to represent the historical modes, resulting in "catastrophic forgetting." To address these problems, this article proposed a jointly mode-matching and similarity-preserving dictionary learning (JMSDL) method, which updated the model by learning the data of new modes, so that the model can adaptively match the newly emerged modes. At the same time, a similarity metric was put forward to guarantee the representation ability of the proposed method for historical data. A numerical simulation experiment, the CSTH process experiment, and an industrial roasting process experiment indicated that the proposed JMSDL method can match new modes while maintaining its performance on the historical modes accurately. In addition, the proposed method significantly outperforms the state-of-the-art methods in terms of fault detection and false alarm rate.
In this paper, a novel variational inference semi-supervised GMM (VI-S(2)GMM) model is firstly proposed for multimodeprocess predictive modeling with semi-supervised data. Since all the labeled and unlabeled data sam...
详细信息
ISBN:
(纸本)9781538626184
In this paper, a novel variational inference semi-supervised GMM (VI-S(2)GMM) model is firstly proposed for multimodeprocess predictive modeling with semi-supervised data. Since all the labeled and unlabeled data samples are involved in each iteration of parameter updating, an intractable computing problem occurs when facing a high-dimension and large-scale dataset. To tack this problem, a scalable Stochastic Variational Inference semi-supervised GMM (SVI-S(2)GMM) is further proposed for massive semi-supervised data. Through taking advantage of stochastic gradient optimization algorithm to maximize the Evidence of Lower Bound (ELBO), the VI-based algorithm becomes scalable. In SVI-S(2)GMM, only one or a mini-batch of samples is randomly selected to update parameters in each iteration, which is more efficient than VI-S(2)GMM. In this way, a large number of unlabeled process data can be useful in the modeling, which will benefit the parameter identification. The SVI-S(2)GMM is then exploited for the prediction of quality-related key performance index (KPI). Two modeling cases with large scale of semi-supervised datasets demonstrate the feasibility and effectiveness of the proposed algorithms.
Finite Gaussian mixture model (GMM) has recently proven to be a powerful unsupervised treatment for monitoring nonlinear processes with multiple operating conditions. The performance of GMM-based monitoring method lar...
详细信息
Finite Gaussian mixture model (GMM) has recently proven to be a powerful unsupervised treatment for monitoring nonlinear processes with multiple operating conditions. The performance of GMM-based monitoring method largely depends on the number of mixture densities. However, the popular penalty method, such as Bayesian information criterion (BIC) and Akaike's information criterion (AIC), usually tend to yield noisy model size estimates. Moreover, the parameter estimates in GMM are susceptible to outliers. To overcome these deficiencies, this paper proposes a new process monitoring technique based on a robust infinite Gaussian mixture model (Ro-IGMM). Specifically, a separate weight at each point is assigned to the precisions as a measure of smoothness, representing the similarities to other data points. The Chinese restaurant process is then placed on a prior to turn into infinite groupings. The informations, such as a distribution over the number of clusters, the cluster assignments, and the parameters associated with each cluster, can be given by the posterior which is obtained by a collapse Markov chain Monte Carlo (MCMC) inference. Simulation results on the benchmark Tennessee Eastman process show that Ro-IGMM-based process monitoring method is more insensitive to outliers during processmodeling, compared to traditional methods working with BIC model selection.
作者:
Yao, LeGe, ZhiqiangZhejiang Univ
Coll Control Sci & Engn State Key Lab Ind Control Technol Hangzhou 310027 Zhejiang Peoples R China Minist Educ
Key Lab Knowledge Automat Ind Proc Beijing 100083 Peoples R China
In this paper, a novel variational inference semisupervised Gaussian mixture model (VI-S-2 GMM) model is first proposed for semisupervised predictive modeling in multimodeprocesses. Parameters of Gaussian components ...
详细信息
In this paper, a novel variational inference semisupervised Gaussian mixture model (VI-S-2 GMM) model is first proposed for semisupervised predictive modeling in multimodeprocesses. Parameters of Gaussian components are identified more accurately with extra unlabeled samples, which improve the prediction performance of the regression model. Since all labeled and unlabeled data samples are involved in each iteration of parameter updating, intractable computing problems occur when facing high-dimension datasets. To tackle this problem, a scalable stochastic VI-S-2 GMM (SVI-S-2 GMM) is further proposed. Through taking advantage of a stochastic gradient optimization algorithm to maximize the evidence of lower bound, the VI-based algorithm becomes scalable. In the SVI-S-2 GMM, only one or a minibatch of samples is randomly selected to update parameters in each iteration, which is more efficient than the VI-S-2 GMM. Since the whole dataset is divided and transferred to iterations batch by batch, the scalable SVI-S-2 GMM algorithm can easily handle the big data modeling issue. In this way, a large number of unlabeled data can be useful in the modeling, which will further benefit the prediction performance. The SVI-S-2 GMM is then exploited for the prediction of a quality-related key performance index. Two examples demonstrate the feasibility and effectiveness of the proposed algorithms.
暂无评论