Original Process Data Research Articles

Adaptive denoising autoencoder for robust fault detection

In real chemical processes, the collected data is often subject to interference from ambient environmental noise, resulting in a decline in the detection performance. Although denoising autoencoder can realize robust fault detection to a certain extent, its performance is limited by the addition of artificial noise, which refers to Gaussian noise or the dropout noise. Moreover, the artificial noise added is often far from the distribution of real data, making it difficult to suppress the noise in real data. To address this issue, an adaptive denoising autoencoder (ADAE) based on the diffusion model is proposed. ADAE adopts a chain-like structure, concatenating traditional DAEs by replacing the manually injected noise in each DAE block with noise generated by diffusion model at various diffusion steps. In order to better capture the key features of the original process data, a novel noise prediction network is proposed within the diffusion model, utilizing dilated and gated convolutions to capture the dynamic information of the process data. Ultimately, adaptation is achieved through a chain of DAEs, suppressing noise similar to the distribution of the original data. The superiority of ADAE in noise suppression is then verified through numerical examples and the Tennessee Eastman process (TEP), comparing it with other methods.

A multi-feature-based fault diagnosis method based on the weighted timeliness broad learning system

Accurate and timely fault diagnosis is a vital task to ensure process safety of modern industrial facilities. Motivated by the complex variations of process signals and mutual coupling of faults, this paper presents a multi-feature-based fault diagnosis method based on the weighted timeliness Broad Learning System (BLS). The proposed method fuses multiple features extracted from the original process data to improve the fault diagnosis performance, and makes the diagnosis model suitable for dynamic fault diagnosis problems by incorporating the BLS. The major contributions of this study are twofolds: 1) A systematic multi-feature extraction method is proposed to extract long-term trend features, short-term trend features, and binary alarm signals, which reflect the direction and amplitude changes of process signals under faulty conditions; 2) a weighted timeliness BLS structure with multiple fault-sensitive features as the input is proposed to ensure the dynamic characteristics of the fault diagnosis model. The designed fault diagnosis model can be updated in an incremental manner, and thus can improve the model updating efficiency while ensuring accuracy. The effectiveness and superiority of the proposed method is demonstrated by a case study based on the Tennessee Eastman benchmark process.

Distributed status evaluation method based on evidence revision fusion

BackgroundComplex process industrial process is large in scale, with numerous variables and complex variable relationship. It is difficult to construct an overall status evaluation model to accurately evaluate the operation status. Compared with holistic modeling status evaluation, distributed status evaluation can make better use of potentially valuable local details. Besides, the integration of result from multiple status evaluation models using decision fusion method such as Dempster-Shafer (DS) evidence theory can make the model more robust. However, DS evidence theory produces counterintuitive result when dealing with high conflict evidence. MethodsIn this paper, a distributed status evaluation method based on evidence revision fusion is proposed. Firstly, a partition method combining qualitative and quantitative knowledge is proposed, which improves the ability to capture local effective information. Secondly, a new status fusion evaluation method is proposed. Feature fusion technique is used to process the original process data, and the evaluation results of multiple status evaluation models are fused by the decision fusion method. Finally, a conflict management strategy with parallel supervised and unsupervised revision is proposed to revise high-conflict evidence from different status evaluation models. FindingsThe proposed method shows good performance in Tennessee-Eastman process, which demonstrates the effectiveness of the method.

Predictive Modeling With Multiresolution Pyramid VAE and Industrial Soft Sensor Applications.

In industrial processes, the sampling rates of process variables are discrepant because of the nature of instruments and measuring demands, which forms the challenging issue, that is, the multirate modeling in the data-driven soft sensor development. In this work, a multiresolution pyramid variational autoencoder (MR-PVAE) predictive model is proposed to solve this problem based on the deep feature extraction and feature pyramid augmentation. First, a multirate data filter is designed through a resolution searching strategy to turn the original process data into a multiresolution dataset. Then, the pyramid variational autoencoder (PVAE) is proposed to extract deep nonlinear features from the data with different resolutions. In PVAE, the augmented feature pyramid is constructed layer by layer to fuse extracted features from low resolution to the high. As a consequence, the extracted features with various resolutions are gathered to form the regression model, where the process information contained in data with discrepant sampling rates can be fully utilized. Due to the layer-by-layer enhanced features, the prediction accuracy of the soft sensing model are gradually improved. Meanwhile, an optimized training strategy is established to select the optimal feature pyramid for prediction. A numerical experiment and an industrial soft sensing case are given to validate the effectiveness and superiority of the proposed MR-PVAE model.

Dynamic convolutional gated recurrent unit attention auto-encoder for feature learning and fault detection in dynamic industrial processes

The dynamic characteristics (i.e. autocorrelation and cross-correlation) in modern industrial systems have raised great challenges to process fault detection. To cope with the dynamics and uncertainty of dynamic industrial processes, this paper proposes a new process control method, dynamic convolutional gated recurrent unit attention auto-encoder (DCGRUA-AE) for fault detection in dynamic processes. Firstly, DCGRUA-AE integrates a convolutional gated recurrent unit (CGRU) with a local convolution layer to learn both global and local features of dynamic process data in an unsupervised fashion. Secondly, a dual attention module is embedded in the deep network to preserve effective features. Finally, DeconvGRU combined with a dense layer is used as the encoder to reconstruct the original process data. Two statistics (i.e. T-square (T 2) and squared prediction error (SPE)) based on DCGRUA-AE are used to set the control limits for fault detection. The feasibility and superiority of DCGRUA-AE-based fault detection method have been verified on four industrial processes. The experimental results indicate that the CGRU and dual attention mechanism can significantly improve the fault detection performance of DCGRUA-AE in dynamic processes. The hybrid of CGRU, attention mechanism and auto-encoder provides a new method for fault detection in dynamic industrial processes.

Deep Canonical Correlation Analysis Using Sparsity-Constrained Optimization for Nonlinear Process Monitoring

This article proposes an efficient nonlinear process monitoring method (DCCA-SCO) by integrating canonical correlation analysis (CCA), deep autoencoder neural networks (DAENNs), and sparsity-constrained optimization (SCO). Specifically, DAENNs are first used to learn a nonlinear function automatically, which characterizes intrinsic features of the original process data. Then, the CCA is performed in that low-dimensional representation space to extract the most correlated variables. In addition, the SCO is imposed to reduce the redundancy of the hidden representation. Unlike other deep CCA methods, the DCCA-SCO provides a new nonlinear method that is able to learn a nonlinear mapping with a sparse prior. The validity of the proposed DCCA-SCO is extensively demonstrated on the benchmark Tennessee Eastman (TE) process and the diesel generator process. In particular, compared with the classical CCA, the fault detection rate is increased by 8.00% for the fault IDV(11) in the TE process.

Cooperative Deep Dynamic Feature Extraction and Variable Time-Delay Estimation for Industrial Quality Prediction

In this article, a novel data-driven industrial quality predictor is proposed based on the cooperative deep dynamic feature extraction and variable time-delay (VTD) estimation. A semisupervised dynamic feature extracting (SSDFE) network is first proposed to extract nonlinear dynamic features to build a regression model for output quality prediction. Due to the inherent process structure and different positions of sampling instruments, time-delays commonly exist between process variables and quality variables, which may distort the original distribution and relationship in collected data. To recover the original process data pattern, the VTDs are regarded as model parameters and cooperatively obtained in the training process of the SSDFE network through an integer differential evolution algorithm. With the estimated VTD values, the reconstructed dataset further helps improve the prediction performance of the proposed SSDFE network. Two case studies are presented to demonstrate the superiority of the proposed method with VTD estimation.

Fault diagnosis of microbial pharmaceutical fermentation process with non-Gaussian and nonlinear coexistence

A large Proportion of batch processes commonly have traits of non-Gaussian and nonlinear. In this work, Multiway Kernel Entropy Independent Component Analysis (MKEICA) algorithm was developed to formulate more accurate model for process monitoring so as to enhance the monitoring performance. The original process data with three-dimension were first expanded into two-dimensional data matrix by using AT variable expansion method. The Kernel Entropy Component Analysis (KECA) was then employed to preprocess the data in order to reduce data redundancy. Such approach can also retain the information of cluster structure and maximize the essential characteristics of data. After that, a monitoring model of MKEICA was established for production process monitoring. Once a fault is detected, a nonlinear contribution plots method would be utilized to diagnose the fault variables. Consequently, to illustrate the superiority and feasibility, the proposed method was conducted on the penicillin simulation platform and the actual pharmaceutical production process.

Novel Pattern-Matching Integrated KCVA with Adaptive Rank-Order Morphological Filter and Its Application to Fault Diagnosis

With the scale expansion of industrial processes, the relationship between process variables becomes complex and highly nonlinear. As a result, the requirements for fault diagnosis and safety monitoring become demanding. To address this problem, a novel and effective pattern matching method using kernel canonical variate analysis (KCVA) integrated with an adaptive rank-order morphological filter (ARMF) is proposed for fault diagnosis. In the proposed method, KCVA is first used to extract the nonlinear correlation information with dynamic characteristics from the original process data and achieve feature dimension reduction; the features extracted by KCVA are then subjected to ARMF transformation for output trend features and pattern matching. To accurately evaluate the morphological similarity between the test trends and template trends of ARMF, the dynamic time warping distance is adopted for pattern classification. Finally, the proposed KCVA-ARMF pattern matching method is developed as an effective faul...

Data Augmentation Applied to Machine Learning-Based Monitoring of a Pulp and Paper Process

Industrial archived process data represent a convenient source of information for data-driven models, such as artificial neural network (ANN), that can be used for safety and efficiency improvement like early or even predictive fault detection and diagnosis (FDD). Nonetheless, most of the data used for model generation are representative of the process nominal states and therefore are not enough for classification problems intended to determine abnormal process conditions. This work proposes the use of techniques to augment the original real data standards, dismissing the need for experiments that could jeopardize process safety. It uses the Monte Carlo technique to artificially increase the number of model inputs coupled to the nearest neighbor search (NNS) by geometric distances to consistently classify the generated patterns in normal or faulty statuses. Finally, a radial basis function neural network is trained with the augmented data. The methodology was validated by a study case in which 3381 pulp and paper industrial data points were expanded to monitor the formation of particles in a recovery boiler. Only 5.8% of the original process data were examples of faulty conditions, but the new expanded and balanced data collection leveraged the classification performance of the neural network, allowing its future use for monitoring purpose.

A time-delay analysis method for the variables of grinding process

Data based expert knowledge mining is very important for intelligent and optimal grinding process control. However, the original process data includes noise, and there is a certain time delay between the variables of grinding process. These factors bring some difficulties to the extraction of expert knowledge based on data analysis. To solve this problem, a method of combining wavelet de-noising with cross-correlation is proposed for the time-delay analysis of variables in grinding process. The case study on the grinding process shows the effectiveness of the proposed method.

Development of a multi-model strategy based soft sensor using gaussian process regression and principal component analysis in fermentation processes

In fermentation processes, single model based soft sensors cannot guarantee prediction performance owing to process characteristics of non-linearity, shifting operating modes, dynamics and uncertainty. In this paper, a novel multi-model based modeling method using Gaussian process regression (GPR) and principal component analysis (PCA) was proposed to construct a soft sensor for biomass concentration estimation in fermentation processes. In the method, principal components (PCs) extracted from original process data are firstly used to build GPR based sub-models. Then, to obtain final predictions, posteriori probabilities of the GPR based sub-models are used to combine outputs of sub-models. The proposed soft sensor was validated on simulation data of a Penicillin fermentation process. For comparisons, several other soft sensor models, e.g. GPR, back-propagation neural network (BP-NN) and least square support vector machine (LSSVM), were also studied. Results show that the proposed soft sensor has better prediction accuracy and smaller confidence intervals.

Unevenly Sampled Dynamic Data Modeling and Monitoring With an Industrial Application

In this paper, a dynamic modeling method for unevenly sampled data is proposed for the monitoring of bi-layer (i.e., a process layer and a quality layer) dynamic processes. First, a novel uneven data dynamic canonical correlation analysis method with an integrated dynamic time window is proposed for interlayer latent structure modeling, which captures the dynamic relations between regularly sampled process data and quality data with slow and irregular sampling. The new model is a step toward big data modeling to deal with data irregularity and diversity. Second, after extracting covariations using an interlayer model, intralayer variations are extracted using subsequent principal component analysis on the residual subspaces of the original process data and quality data, respectively. Third, a concurrent monitoring method for unevenly sampled bi-layer data is proposed. Finally, the proposed method is demonstrated using an illustrative simulation example and applied successfully to a real blast furnace iron-making process.

A Kernel Time Structure Independent Component Analysis Method for Nonlinear Process Monitoring

Kernel independent component analysis (KICA) is a newly emerging nonlinear process monitoring method, which can extract mutually independent latent variables called independent components (ICs) from process variables. However, when more than one IC have Gaussian distribution, it cannot extract the IC feature effectively and thus its monitoring performance will be degraded drastically. To solve such a problem, a kernel time structure independent component analysis (KTSICA) method is proposed for monitoring nonlinear process in this paper. The original process data are mapped into a feature space nonlinearly and then the whitened data are calculated in the feature space by the kernel trick. Subsequently, a time structure independent component analysis algorithm, which has no requirement for the distribution of ICs, is proposed to extract the IC feature. Finally, two monitoring statistics are built to detect process faults. When some fault is detected, a nonlinear fault identification method is developed to identify fault variables based on sensitivity analysis. The proposed monitoring method is applied in the Tennessee Eastman benchmark process. Applications demonstrate the superiority of KTSICA over KICA.

Minimum mean squared prediction error criterion based improved independent component analysis method for process monitoring

Independent component analysis (ICA) is a newly emerging feature extraction method for non-Gaussian process monitoring. However, the extracted feature by ICA may not represent the original process data well, which can result in the degraded monitoring performance. In this paper, an improved ICA method based on the minimum mean squared prediction error criterion is proposed for process monitoring. A new criterion which can make the extracted non-Gaussian feature be efficient representation for the original process data is constructed as the objective function of the improved ICA by integrating the maximum negentropy criterion of the conventional ICA with the minimum mean squared prediction error criterion. Then the gradient ascent algorithm is applied to optimize the constructed objective function for seeking the feature extraction directions. Finally, a monitoring statistic is built based on the extracted feature to detect process faults. The simulation studies on the Tennessee Eastman benchmark process demonstrate that the improve ICA is more effective than the conventional ICA for improving the monitoring performance in terms of the fault detection rate.

Improved kernel PCA-based monitoring approach for nonlinear processes

Conventional kernel principal component analysis (KPCA) may not function well for nonlinear processes, since the Gaussian assumption of the method may be violated through nonlinear and kernel transformation of the original process data. To overcome this deficiency, a statistical local approach is incorporated into KPCA. Through this method, a new score variable which was called improved residual in the statistical local approach is constructed. The new variable approximately follows Gaussian distribution, in spite of which distribution the original data follows. Two new statistics are constructed for process monitoring, with their corresponding confidence limits determined by a χ 2 distribution. Besides of the improvement made on KPCA, the new joint local approach-KPCA method also shows superiority on detection sensitivity, especially for small faults slow changes of the process. The new method is exemplified using a numerical study and also tested in the complicated Tennessee Eastman (TE) benchmark process.

Development of a Soft Sensor for a Batch Distillation Column Using Support Vector Regression Techniques

A support vector regression (SVR)-based model is developed for a batch distillation process in order to estimate the product compositions from temperature measurements. Kernel function such as linear, polynomial and RBF are employed for SVR modelling. The original process data was generated by simulating the batch distillation process, varying the initial feed composition and boilup rate from batch to batch. Within each batch reflux ratio was also randomly changed to represent the true dynamics of the batch distillation. The results show the potential of the method for developing softsensor for chemical processes.

Process monitoring based on classification tree and discriminant analysis

To cope with the computational intensity associated with classification tree analysis and the multicolinearity in the process data, a newly developed process monitoring scheme integrating classification tree and Fisher Discriminant Analysis (FDA) is developed. FDA extracts the most significant components in the original process data and achieves optimal discriminating among different faults. Classification tree uses the FDA scores, which are the lower dimensional representation produced by FDA, to separate observations into different fault classes. A stopping rule is applied to determine the optimal order of FDA. Two case studies are presented to illustrate the effectiveness of the proposed methods compared with the original classification tree. The new method generates better classification accuracy and uses less construction time.

DEVELOPMENT OF A SOFT SENSOR FOR A BATCH DISTILLATION COLUMN USING LINEAR AND NONLINEAR PLS REGRESSION TECHNIQUES

A PLS-based model is developed for a batch distillation process in order to estimate the product compositions from temperature measurements. Both linear and nonlinear versions of PLS are employed and their estimation performance is compared. Several issues are addressed such as the selection of the most appropriate model input variables, and the effect of augmenting the original process data with lagged measurements. A novel PLS approach is also proposed that provides for the development of multiple PLS models for different time intervals during the batch operation.