Recursive Gaussian Mixture Models for Adaptive Process Monitoring

Zheng, Junhua; Wen, Qiaojun; Song, Zhihuan

doi:10.1021/acs.iecr.8b06101

Cited by 24 publications

(7 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Future directions will be devoted to the extension of the proposed approach to distributed and dynamic scenarios (e.g., pan-tilt or mobile cameras); in particular, incremental GMM approaches may be used to update and refine the SM in time ( [56]), while concepts of federated machine learning may be applied to distribute the computation over the network [57]. In addition, it is important to formally study the impact of the main setup parameters on the overall identification performance, as suggested by the results in Section 6.1.…”

Section: Discussionmentioning

confidence: 99%

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

Varotto,

Fabris,

Michieletto

et al. 2022

Preprint

View full text Add to dashboard Cite

Visual sensor networks constitute a fundamental class of distributed sensing systems, with unique complexity and performance research subjects. One of these novel challenges is represented by the identification of the network stimulation model (SM), which emerges when a set of detectable events trigger different subsets of the cameras. In this direction, the formulation of the related SM identification problem is proposed, along with a proper network observations generation method. Consequently, an approach based on deep embedded features and soft clustering is leveraged to solve the presented identification problem. In detail, the Gaussian Mixture Modeling is employed to provide a suitable description for data distribution and an autoencoder is used to reduce undesired effects due to the so-called curse of dimensionality. Hence, it is shown that a SM can be learnt by solving Maximum A-Posteriori estimation on the encoded features belonging to a space with lower dimensionality. Lastly, numerical results are reported to validate the devised estimation algorithm.

show abstract

Section: Discussionmentioning

confidence: 99%

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

Varotto,

Fabris,

Michieletto

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

Section: Data Types and Data Transformation Tomentioning

confidence: 99%

“…The probability density function (PDF) of the measurement errors is estimated to quantify the range of the measurement errors and to evaluate the uncertainty that is caused by the measurement errors . Since the PDF of the measurement errors may not follow a Gaussian distribution, a Gaussian mixture model (GMM), , which can approximate a wide variety of PDFs, is used in this work. The PDF of the measurement errors of the j th variable is expressed as where M is the number of Gaussian components; ω m denotes the weight of the m th component and satisfies 0 ≤ ω m ≤ 1 and ; θ m = { μ m ,Σ m } consists of the mean μ m and the covariance Σ m and represents the density function parameter set; and p ( e j | θ m ) denotes the m thGaussian PDF, which is expressed as follows: As shown in Figure , the upper bound and the lower bound of the measurement error can be determined as the upper and lower quantiles, respectively, with respect to the significance α: where and refer to the upper bound and the lower bound, respectively, of the measurement error for the j th variable.…”

Section: Data Types and Data Transformation To Interval Formatmentioning

confidence: 99%

Abnormal Condition Identification via OVR-IRBF-NN for the Process Industry with Imprecise Data and Semantic Information

Yuan

Wang

et al. 2020

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Data-driven methods are commonly used to identify abnormal operating conditions to maintain the health and safety of processes, which assume that the process data are precisely known and single-valued. However, in practice, process data are from multiple sources and are in various formats with uncertainties or measurement errors, which may lead to a high false-alarm rate and imprecise decisions. In addition, various key variables are difficult or impossible to measure online, and they are always estimated and described in terms of semantic information by operators or experts. In this work, a one-versus-rest interval radial basis function neural network (OVR-IRBF-NN)-based abnormality identification method is proposed for imprecise data and semantic information in the process industry. First, three types of process data, namely, precise single-valued variables (PSVVs), imprecise single-valued variables (ISVVs), and two-dimensional-interval-valued variables (2DIVVs), which are transformed from semantic information, are analyzed to investigate more comprehensive process information. Then, transformation approaches are presented for transforming these three types of data into one-dimensional intervals. Gaussian mixture model (GMM)-based measurement error estimation is developed for ISVVs, and an uncertain-unchanged interval adjustment strategy is proposed for 2DIVVs. Moreover, the one-versus-rest interval radial basis function neural network (OVR-IRBF-NN) is put forth for the identification of abnormal operating conditions via the analysis of one-dimensional-interval-valued data, providing more reliable and robust identification results due to the employment of interval-valued data. The proposed strategy is evaluated on both a numerical example and a real hydrometallurgical metallurgy leaching process, and the results demonstrate the feasibility and effectiveness of the strategy.

show abstract

“…For the numerical study, the proposed method is investigated by the typical nonlinear sinc function. Then the method is applied to estimate the particle size in a cobalt oxalate synthesis pilot (26) where N denotes the number of test samples, y n andŷ n denote the measured value and the predicted value by the soft sensor model respectively.…”

Section: Case Studiesmentioning

confidence: 99%

“…JITL method is an alternative solution to deal with abrupt changes in process behaviors. Moreover, all the three updating strategies are combined with the traditional soft sensors, although a recursive GMM and a moving window GMM method were proposed in [26], [27] respectively. However, the two approaches were applied to process modeling and monitoring for dynamic multimode processes.…”

Section: Introductionmentioning

confidence: 99%

Soft Sensor Model Development for Cobalt Oxalate Synthesis Process Based on Adaptive Gaussian Mixture Regression

et al. 2019

View full text Add to dashboard Cite

Average particle size is a critical quality variable for optimal operation of the cobalt oxalate synthesis process. However, the measurement for this variable is often achieved from offline laboratory assaying procedure with low sampling rate and low reliability. Therefore, a data-driven soft sensor model based on adaptive Gaussian mixture regression (AGMR) is presented in this paper. Firstly, a GMR based soft sensor model is developed for predicting the average particle size. Secondly, the prediction uncertainty obtained from the GMR model is used to assess the performance of the current soft sensor model. Thirdly, a dual updating algorithm based on the model performance assessment is constructed to track the timevarying behavior of the synthesis process. In the updating method, bias updating and moving window model updating methods are performed in turns based on the results of model performance assessment. The dual updating mechanism can avoid blind updating. Finally, a numerical example and a real industrial cobalt oxalate synthesis process application are used to demonstrate the effectiveness of the proposed method. INDEX TERMS Cobalt oxalate synthesis process, Gaussian mixture regression, soft sensor, model performance assessment, model updating.

show abstract

Recursive Gaussian Mixture Models for Adaptive Process Monitoring

Cited by 24 publications

References 44 publications

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

Visual Sensor Network Stimulation Model Identification via Gaussian Mixture Model and Deep Embedded Features

Abnormal Condition Identification via OVR-IRBF-NN for the Process Industry with Imprecise Data and Semantic Information

Soft Sensor Model Development for Cobalt Oxalate Synthesis Process Based on Adaptive Gaussian Mixture Regression

Contact Info

Product

Resources

About