Fault Detection In Industrial Processes Research Articles

Industrial process fault detection and diagnosis framework based on enhanced supervised kernel entropy component analysis

Most existing industrial process fault detection and diagnosis (FDD) techniques operate on data collected at a single scale and focus only on known faults. However, actual process data are inherently multiscale and unknown faults are always inevitable during system running. Therefore, they may perform unsatisfactorily. To tackle this problem, this paper develops a decentralized industrial process FDD framework using multiple enhanced supervised kernel entropy component analysis (enhanced SKECA) models, where each model acts as a fault indicator for one specific fault. Faults can be easily diagnosed by monitoring the outputs of all models within the framework. In particular, when new faults are identified, the framework can update itself only by adding the corresponding enhanced SKECA models without a complete rebuilding process. The monitoring results for the continuous stirred tank reactor (CSTR) process show that the proposed framework is effective in diagnosing both known and unknown faults.

Critical Concurrent Feature Selection and Enhanced Heterogeneous Ensemble Learning Approach for Fault Detection in Industrial Processes

Machine learning approaches have been successfully applied to fault detection in complex industrial processes, while it is still challenging to design an optimal feature space and a stable model for various faults. To solve this problem, a hybrid approach with critical concurrent feature selection and enhanced heterogeneous ensemble learning is proposed. First, raw process variables from field instruments and slow features obtained by simple linear transformation are ingeniously combined to construct the concurrent feature space so as to contain static and dynamic information. Then, the critical concurrent feature space of each individual learner is automatically obtained by the optimal selection process, and the feature diversity of ensemble is realized simultaneously. Finally, an enhanced heterogeneous ensemble model is constructed by different optimized individual learners, which effectively improves the classification accuracy and stability in fault detection. The performance of the proposed approach is evaluated in a simulated Tennessee Eastman benchmark and a real-word three-phase flow process. The experimental results illustrate that the performance of each optimized individual learner outperforms classic random forest algorithm. The generalization and stability of the ensemble model are further improved. Compared with traditional classification algorithms, the proposed approach achieves superior performance with accuracy that exceeds 89% for the Tennessee Eastman process and exceeds 88% for the three-phase flow process respectively. Additionally, the selected critical concurrent features indicate that both static and dynamic information play important roles in fault detection of industrial processes.

Adversarial Autoencoder Based Feature Learning for Fault Detection in Industrial Processes

Deep learning has recently emerged as a promising method for nonlinear process monitoring. However, ensuring that the features from process variables have representative information of the high-dimensional process data remains a challenge. In this study, we propose an adversarial autoencoder (AAE) based process monitoring system. AAE which combines the advantages of a variational autoencoder and a generative adversarial network enables the generation of features that follow the designed prior distribution. By employing the AAE model, features that have informative manifolds of the original data are obtained. These features are used for constructing and monitoring statistics and improve the stability and reliability of fault detection. Extracted features help calculate the degree of abnormalities in process variables more robustly and indicate the type of fault information they imply. Finally, our proposed method is testified using the Tennessee Eastman benchmark process in terms of fault detection rate, false alarm rate, and fault detection delays.

Improving Fault Detection in Industrial Processes by Event-Driven Data Acquisition

Data acquisition in process industries usually takes place at each sampling. The disadvantage is that a considerable amount of data without new information about the state of the process is continuously transmitted and processed. This negatively affects the communication system and computational power, which is more critical nowadays given the number of variables measured, even in seconds. One solution concerns the event-driven paradigm, in which only relevant data according to a pre-defined criterion is forwarded for further processing. This work investigated the event-based threshold and delta methods in the context of fault detection. The data transmission rate was also analyzed. The well-know Tennessee Eastman problem (TEP) was used as a case study. The fault detection system was based on PCA (principal component analysis), which is widely used for this purpose in this benchmark. The results were compared with the commonly used time-based approach, for a fixed false alarm rate. The threshold rule provided similar results, but with much less data. For the delta rule, significant MDR (missed detection rate) gains of up to 74% were obtained for five of the six hard-to-detect faults, and of up to 69%, for two of the three very hard-to-detect faults. MDR values very close to zero were reached for two of the three intermediate detection faults and two of the hard-to-detect faults. The detection time was also evaluated. In this regard, considerably lower values were obtained for all intermediate detection faults, three of the hard-to-detect faults and all very hard-to-detect faults. In short, the delta method was able to improve fault detection performance, especially for hard-to-detect faults, with a considerably lower data transmission rate, around 20% on average. Event-driven data acquisition can be very attractive for process industries.

Industrial Process Fault Detection Based on Entropy Score Contribution Analysis

Aiming at the problem of poor fault detection ability in complex industrial processes, this paper proposes a fault detection method of entropy score contribution analysis, which can better determine the number of samples with a high contribution rate and achieve good fault detection results. First, normalize the input data to obtain the Euclidean distance of each line vector; Secondly, the number of samples with a high contribution rate is determined by entropy score contribution and determine the initial threshold; Thirdly, a parameter adaptive strategy is proposed to set the threshold of statistics; Finally, fault detection is carried out on the statistical data, and the size of fault detection rate and false alarm rate is calculated. In the fourth part of this paper, we use this fault detection method to detect the faults from the Tennessee Eastman process. The simulation results show that the method proposed has more advantages than the traditional methods, and can be effectively applied to complex industrial processes.

Visual Analytics: A New Paradigm for Process Monitoring

As a result of recent breakthroughs in computer vision technologies, significant research interest has emerged to encode process data into visual clues and treat process monitoring problems as computer vision tasks. Imaging time-series signals as a feature engineering step forms a new branch of data analytics called ”visual analytics”. In the context of process monitoring, we define visual analytics as the integration of visual representation of the data combined with the use of computer vision tools and analytical reasoning to support decision-making and knowledge extraction from the data. In this work, a novel end-to-end visual analytics pipeline for industrial process fault detection using 1D and 2D convolution operations is proposed. The proposed approach presents a visual representation of data that captures temporal and local features from historical time-series signals. Next, the learned features in a 2D format are visually recognized and classified using 2D convolution operations. Our experimental results demonstrate that this approach achieves better performance on an industrial multivariate dataset compared to other state-of-art signals imaging tools such as Gramian Angular Field (GAF) and Recurrence Plots (RP).

Unstructured borderline self-organizing map: Learning highly imbalanced, high-dimensional datasets for fault detection

Fault detection in industrial processes is critical for yield improvement and manufacturing cost reduction. However, most industrial processes produce highly imbalanced and high-dimensional datasets, in which the normal data overwhelm the fault data in number and many noninformative features add noise to the data distribution. Thus, addressing class imbalance and high-dimensionality problems has been considered key to successful fault detection. In this paper, we propose a novel model called an unstructured borderline self-organizing map (UB-SOM) designed to solve these two problems. UB-SOM not only learns the distribution of the normal samples through a small number of representative nodes but also highlights borderline areas. Since UB-SOM yields a new data distribution that emphasizes borderlines, the distributional change from the normal data to the representative nodes reveals which features are considered significant in the borderline areas. We select the significant features based on the featurewise distributional change measured using the Kullback-Leibler divergence. UB-SOM is evaluated based on ten publicly available benchmark imbalanced datasets and two semiconductor process datasets. The experimental results show that we can increase the G-mean by 0.441 for the benchmark datasets and 0.657 for the industrial datasets with data preprocessing throughout UB-SOM. As a result, the proposed method outperforms various undersampling methods incorporating classifier-based feature selection methods.

Industrial process fault detection based on KGLPP model with Cam weighted distance

The nearest neighbor selection of multivariate statistical projection analysis methods assumes locally constant probabilities. However, ignoring the non-uniform distributed characteristic of data causes information redundancy in data-intensive regions and insufficient information in data-sparse regions, leading to detection performance decline. In this study, a new weighted distance named Cam weighted distance is used to reselect the neighbors and consequently overcome the aforementioned limitation. An nonlinear industrial fault detection method based on KGLPP-Cam is developed. The proposed method can preserve not only global and local information but also orientation and adaptive scale to obtain the information of neighbors according to different surroundings. T2 and SPE statistics are calculated for fault detection. A change ratio function is constructed to select sensitive principal components adaptively and better describe the sensitivity of different projection directions for processing change information. The proposed method is examined through a numerical example and TE process.

Echo State Network Based Soft Sensor for Monitoring and Fault Detection of Industrial Processes

In this paper a semi-automatic computationally inexpensive system is developed and implemented for monitoring and fault detection of industrial processes. The system uses a soft sensor based on Echo State Networks (ESN) and is able to capture the non-linear dynamic relationships in the process data, making it convenient for real-time monitoring applications. The soft sensor is set to simulate normal operating conditions, so that when the process is governed by other causes, possibly in failure, high residues occur and allow the failure identification. In addition, the system monitors the reliability of the model predictions by tracking the internal states of the ESN dynamic reservoir, indicating whether the model predictions can be used instead of the measured data. The system is successfully applied to the Mackey-Glass Anomaly Benchmark (MGAB) and to the monitoring of critical pieces of equipment of a real oil and gas plant.

Improved bilayer convolution transfer learning neural network for industrial fault detection

AbstractMachine learning methods have achieved outstanding results in the fault detection of industrial processes when the training and test data follow the same distribution or originate from the same situation. However, the pre‐trained detection model fails when the operation status changes or unknown fault occurs during the actual production. Therefore, this paper proposes a novel bilayer convolutional transfer learning neural network (BCTLNN) to improve the generalization of detection model. BCTLNN is a bilayer network (local and global level) in order to extract the fault features. Transfer learning strategies (fine‐tuning and domain adaptation) are introduced to learn the domain invariant features by minimizing the divergence between different domain data. Experiments on the benchmark bearing data and the agglomeration fault data from an actual polyethylene process are employed to verify the effectiveness of the proposed method.

Cost-benefit analysis of calibration model maintenance strategies for process monitoring

The long-term prediction performance of spectroscopic calibration models is a critical factor to monitor or control many production processes. Over time, new variations may emerge that deteriorate prediction performance. Therefore, models have to be maintained to retain or improve their prediction performance through time, requiring considerable resources and data. Maintenance should improve relevant predictions but also needs to be resource and cost efficient. Current approaches do not consider these trade-offs. We propose a new method to quantify the effectiveness and cost of model maintenance strategies based on historical data. Model performance over time for past, imminent and future samples is evaluated as these may react differently to maintenance. The model performance and required updating resources are translated into relative cost and benefit to compare strategies and determine optimal maintenance parameters. We used this method to evaluate a maintenance strategy that combines adding incoming samples to the calibration data with re-optimization of spectral preprocessing and modelling parameters. Continuously adding samples to the calibration data is shown to improve prediction performance and leads to more robust and generic models for emerging variations in all investigated data streams. Selectively adding incoming sample variations showed a reduced prediction performance but saves considerably in resources. Comparing model performance on the different sampling windows can also be used to determine an optimal updating frequency. This novel strategy to evaluate the expected performance and determine an optimal maintenance strategy is generally applicable and should lead to robust and consistently high prospective and/or retrospective model performance through time, which can be crucial for optimal operation and fault detection in industrial processes.

Change point detection for compositional multivariate data

Change point detection algorithms have numerous applications in areas of medical condition monitoring, fault detection in industrial processes, human activity analysis, climate change detection, and speech recognition. We consider the problem of change point detection on compositional multivariate data (each sample is a probability mass function), which is a practically important sub-class of general multivariate data. While the problem of change-point detection is well studied in univariate setting, and there are few viable implementations for a general multivariate data, the existing methods do not perform well on compositional data. In this paper, we propose a parametric approach for change point detection in compositional data. Moreover, using simple transformations on data, we extend our approach to handle any general multivariate data. Experimentally, we show that our method performs significantly better on compositional data and is competitive on general data compared to the available state of the art implementations.

Industrial process fault detection based on deep highly-sensitive feature capture

With the rapid development of sensor and computer technology, deep learning has received extensive attention in the field of fault detection with powerful nonlinear feature extraction capabilities. However, the feature extracted by deep learning contains different fault information volume. Fault detection performance would be affected by features with less fault information. Motivated by this, the deep high-dimensional features extracted from the deep belief network are analyzed, and an index for measuring fault information volume is proposed to select the deep highly-sensitive feature (DHSF) with a large amount of fault information volume as the feature to be detected. Based on DHSFs, Euclidean distance is used for fault detection, and the moving average window function is used to reduce burst noise interference and improve detection performance. Finally, a numerical case and Tennessee Eastman process demonstrate the advantage of the proposed method in fault detection results compared with other methods.

Industrial process fault detection based on locally linear embedded latent mapping

Industrial process fault detection based on locally linear embedded latent mapping

A Novel Mutual Information and Partial Least Squares Approach for Quality-Related and Quality-Unrelated Fault Detection

Partial least squares (PLS) and linear regression methods are widely utilized for quality-related fault detection in industrial processes. Standard PLS decomposes the process variables into principal and residual parts. However, as the principal part still contains many components unrelated to quality, if these components were not removed it could cause many false alarms. Besides, although these components do not affect product quality, they have a great impact on process safety and information about other faults. Removing and discarding these components will lead to a reduction in the detection rate of faults, unrelated to quality. To overcome the drawbacks of Standard PLS, a novel method, MI-PLS (mutual information PLS), is proposed in this paper. The proposed MI-PLS algorithm utilizes mutual information to divide the process variables into selected and residual components, and then uses singular value decomposition (SVD) to further decompose the selected part into quality-related and quality-unrelated components, subsequently constructing quality-related monitoring statistics. To ensure that there is no information loss and that the proposed MI-PLS can be used in quality-related and quality-unrelated fault detection, a principal component analysis (PCA) model is performed on the residual component to obtain its score matrix, which is combined with the quality-unrelated part to obtain the total quality-unrelated monitoring statistics. Finally, the proposed method is applied on a numerical example and Tennessee Eastman process. The proposed MI-PLS has a lower computational load and more robust performance compared with T-PLS and PCR.

Metaheuristics and Support Vector Data Description for Fault Detection in Industrial Processes

In this study, a system for faults detection using a combination of Support Vector Data Description (SVDD) with metaheuristic algorithms is presented. The presented approach is applied to a real industrial process where the set of measured faults is scarce. The original contribution in this work is the industrial context of application and the comparison of swarm intelligence algorithms to optimize the SVDD hyper-parameters. Four recent metaheuristics are compared hereby to solve the corresponding optimization problem in an efficient manner. These optimization techniques are then implemented for fault detection in a multivariate industrial process with non-balanced data. The obtained numerical results seem to be promising when the considered optimization techniques are combined with SVDD. In particular, the Spotted Hyena algorithm outperforms other metaheuristics reaching values of F1 score near 100% in fault detection.

A nonlinear method for monitoring industrial process

Aiming at fault detection in industrial processes with nonlinear or high dimensions, a novel method based on locally linear embedding preserve neighborhood for fault detection is proposed in this paper. Locally linear embedding preserve neighborhood is a feature-mapping method that combines Locally linear embedding and Laplacian eigenmaps algorithms. First, two weight matrices are obtained by the Locally linear embedding and Laplacian eigenmaps, respectively. Subsequently, the two weight matrices are combined by a balance factor to obtain the objective function. Locally linear embedding preserve neighborhood method can effectively maintain the characteristics of data in high-dimensional space. The purpose of dimension reduction is to map the high-dimensional data to low-dimensional space by optimizing the objective function. Process monitoring is performed by constructing T2 and Q statistics. To demonstrate its effectiveness and superiority, the proposed locally linear embedding preserve neighborhood for fault detection method is tested under the Swiss Roll dataset and an industrial case study. Compared with traditional fault detection methods, the proposed method in this paper effectively improves the detection rate and reduces the false alarm rate.

Manifold regularized stacked autoencoders-based feature learning for fault detection in industrial processes

Multivariate statistical process control (MSPC) has been widely employed for process fault detection. Recently, deep neural networks (DNNs), i.e., stacked autoencoder (SAE) enjoys its popularization in process fault detection. SAE shows good performance in extracting representative features from the process data based on unsupervised learning, which provides a new monitoring method without large amount of labeled data. However, the extraction of the intrinsic geometrical information from process signals is not considered by these regular SAEs This paper proposes a new DNN, manifold regularized stacked autoencoders (MRSAE) for fault detection in complex industrial processes. The local/global information preservation is incorporated into the encoding phase of SAE to capture intrinsic structure of the process data. MRSAE is used to describe distribution of the nonlinear process data and learn effective features for process fault detection. Two typical statistics (i.e., Hotelling’s T-squared (T2) and squared prediction error (SPE)) based on the extracted features by MRSAE are developed for process fault detection. The comparison between MRSAE and other typical DNNs on a complex numerical process and two benchmark processes, i.e., Tennessee Eastman process (TEP) and Fed-Batch fermentation penicillin process (FBFP), indicates the effectiveness of the proposed method for process fault detection. The manifold regularization-based DNN technique provides a new way for feature learning from high-dimensional and nonlinear process signals.

Graph-Based Method for Fault Detection in the Iron-Making Process

Since the iron-making process is performed in complicated environments and controlled by operators, observation labeling is difficult and time-consuming. Therefore, unsupervised fault detection methods are a promising research topic. Recently, an unsupervised graph-based change point detection method has been introduced, and the graph of observations is constructed by the minimum spanning tree. In this paper, a novel fault detection method based on the graph for an iron-making process is proposed, and a weight calculation method for constructing the minimum spanning tree is introduced. The Euclidean distance and Mahalanobis distance are combined to calculate the weights in the minimum spanning tree, which contain important relations of variables. The distance calculation method is determined by the correlation coefficients of variables. Each testing observation is set as a change point candidate, and a change point candidate divides the observations into two groups. The number of a special type of edge in the minimum spanning tree is used as a fault detection statistic. That special edge connects two observations from two different groups. The minimum number of that type of edge corresponding to the change point candidate is a true change point. Finally, numerical simulation is used to test the power of the proposed method, and a real iron-making process including low stock, cooling, and slip faults is implemented to illustrate the effectiveness of fault detection in industrial processes.

Soft computing applications in the field of industrial and environmental enterprises

Soft computing applications in the field of industrial and environmental enterprises