Process Performance Monitoring Research Articles

Targeted rework strategies for powder bed additive manufacture

A major factor limiting the adoption of powder-bed-fusion additive manufacturing for production of parts is the control of build process defects and the effect these have upon the certification of parts for structural applications. In response to this, new methods for detecting defects and to monitor process performance are being developed. However, effective utilisation of such methods to rework parts in process has yet to be demonstrated.This study investigates the use of spatially resolved acoustic spectroscopy (SRAS) scan data to inform repair strategies within a commercial selective laser melting machine. New methodologies which allow for rework of the most common defects observed in selective laser melting (SLM) manufacturing are proposed and demonstrated. Three rework methodologies are applied to targeted surface breaking pores: a hatch pattern, a spiral pattern and a single shot exposure. The work presented shows that it is possible to correct surface breaking pores using targeted re-melting, reducing the depth of defects whilst minimising changes in local texture. For the hatch rework and spiral rework, a reduction in defect depth of 50% and 31% were observed, respectively, however, no improvement was seen after the single shot exposures. This work is part of a programme to develop a method by which defects can be detected and the part reworked in-process during SLM to enable defect specification targets to be met. Although further work in developing build-characterise-rework strategies for integrated and targeted defect correction is needed, the feasibility of the underlying method of identifying and selectively reworking to reduce a defect has now been demonstrated for the first time.

Near infrared spectroscopy based monitoring of extraction processes of raw material with the help of dynamic predictive modeling

The control of batch-to-batch quality variations remains a challenging task for pharmaceutical industries, e.g., traditional Chinese medicine (TCM) manufacturing. One difficult problem is to produce pharmaceutical products with consistent quality from raw material of large quality variations. In this paper, an integrated methodology combining the near infrared spectroscopy (NIRS) and dynamic predictive modeling is developed for the monitoring and control of the batch extraction process of licorice. With the spectra data in hand, the initial state of the process is firstly estimated with a state-space model to construct a process monitoring strategy for the early detection of variations induced by the initial process inputs such as raw materials. Secondly, the quality property of the end product is predicted at the mid-course during the extraction process with a partial least squares (PLS) model. The batch-end-time (BET) is then adjusted accordingly to minimize the quality variations. In conclusion, our study shows that with the help of the dynamic predictive modeling, NIRS can offer the past and future information of the process, which enables more accurate monitoring and control of process performance and product quality.

A novel process control method for a TT-300 E-Beam/X-Ray system

This paper presents some aspects of the process control method for a TT-300 E-Beam/X-Ray system at Mediscan, Austria. The novelty of the approach is the seamless integration of routine monitoring dosimetry with process data. This allows to calculate a parametric dose for each production unit and consequently a fine grain and holistic process performance monitoring. Process performance is documented in process control charts for the analysis of individual runs as well as historic trending of runs of specific process categories over a specified time range.

Performance monitoring of non-gaussian chemical processes with modes-switching using globality-locality preserving projection

In this paper, we propose a novel performance monitoring and fault detection method, which is based on modified structure analysis and globality and locality preserving (MSAGL) projection, for non-Gaussian processes with multiple operation conditions. By using locality preserving projection to analyze the embedding geometrical manifold and extracting the non-Gaussian features by independent component analysis, MSAGL preserves both the global and local structures of the data simultaneously. Furthermore, the tradeoff parameter of MSAGL is tuned adaptively in order to find the projection direction optimal for revealing the hidden structural information. The validity and effectiveness of this approach are illustrated by applying the proposed technique to the Tennessee Eastman process simulation under multiple operation conditions. The results demonstrate the advantages of the proposed method over conventional eigendecomposition-based monitoring methods.

Evaluation of an electronic nose for odorant and process monitoring of alkaline-stabilized biosolids production

Electronic noses have been widely used in the food industry to monitor process performance and quality control, but use in wastewater and biosolids treatment has not been fully explored. Therefore, we examined the feasibility of an electronic nose to discriminate between treatment conditions of alkaline stabilized biosolids and compared its performance with quantitative analysis of key odorants. Seven lime treatments (0–30% w/w) were prepared and the resultant off-gas was monitored by GC-MS and by an electronic nose equipped with ten metal oxide sensors. A pattern recognition model was created using linear discriminant analysis (LDA) and principal component analysis (PCA) of the electronic nose data. In general, LDA performed better than PCA. LDA showed clear discrimination when single tests were evaluated, but when the full data set was included, discrimination between treatments was reduced. Frequency of accurate recognition was tested by three algorithms with Euclidan and Mahalanobis performing at 81% accuracy and discriminant function analysis at 70%. Concentrations of target compounds by GC-MS were in agreement with those reported in literature and helped to elucidate the behavior of the pattern recognition via comparison of individual sensor responses to different biosolids treatment conditions. Results indicated that the electronic nose can discriminate between lime percentages, thus providing the opportunity to create classes of under-dosed and over-dosed relative to regulatory requirements. Full scale application will require careful evaluation to maintain accuracy under variable process and environmental conditions.

Statistical process monitoring via image data using wavelets

AbstractImage data plays an important role in manufacturing and service industries because each image can provide a huge set of data points in just few seconds with relatively low cost. Enhancement of machine vision systems during the time has led to higher quality images, and the use of statistical methods can help to analyze the data extracted from such images efficiently. It is not efficient from time and cost point of views to use every single pixel in an image to monitor a process or product performance effectively. In recent years, some methods are proposed to deal with image data. These methods are mainly applied for separation of nonconforming items from conforming ones, and they are rarely applied to monitor process capability or performance. In this paper, a nonparametric regression method using wavelet basis function is developed to extract features from gray scale image data. The extracted features are monitored over time to detect process out‐of‐control conditions using a generalized likelihood ratio control chart. The proposed approach can also be applied to find change point and fault location simultaneously. Several numerical examples are used to evaluate performance of the proposed method. Results indicate suitable performance of the proposed method in detecting out‐of‐control conditions and providing precise diagnostic information. Results also illustrate suitable performance of our proposed method in comparison with a competitive approach.

Transforming unstructured natural language descriptions into measurable process performance indicators using Hidden Markov Models

Monitoring process performance is an important means for organizations to identify opportunities to improve their operations. The definition of suitable Process Performance Indicators (PPIs) is a crucial task in this regard. Because PPIs need to be in line with strategic business objectives, the formulation of PPIs is a managerial concern. Managers typically start out to provide relevant indicators in the form of natural language PPI descriptions. Therefore, considerable time and effort have to be invested to transform these descriptions into PPI definitions that can actually be monitored. This work presents an approach that automates this task. The presented approach transforms an unstructured natural language PPI description into a structured notation that is aligned with the implementation underlying a business process. To do so, we combine Hidden Markov Models and semantic matching techniques. A quantitative evaluation on the basis of a data collection obtained from practice demonstrates that our approach works accurately. Therefore, it represents a viable automated alternative to an otherwise laborious manual endeavor.

A Variable-Correlation-Based Sparse Modeling Method for Industrial Process Monitoring

Dimensionality reduction techniques are widely used in data-driven process monitoring methods for extracting key process features from process data. Dimensionality reduction generally leads to information loss, and therefore may reduce the process monitoring performance. However, choosing an appropriate data projection matrix that can minimize the effect of dimensionality reduction on process monitoring performance is often challenging. In this paper, we introduce a method to construct a variable-correlation-based sparse projection matrix (VCBSPM) for reducing the dimension of process data and for building the process monitoring model. The VCBSPM is constructed on the basis of variable correlations, with each column of VCBSPM corresponding to a variable group consisting of highly correlated variables. The VCBSPM has two advantages: (1) it implements dimensionality reduction only for highly correlated variables, and therefore the negative effect of dimensionality reduction on process monitoring performance...

A Novel Decentralized Process Monitoring Scheme Using a Modified Multiblock PCA Algorithm

Decentralized process monitoring based on purely data-based methods has recently gained considerable attention in multivariate statistical process monitoring circle. Although the process variables can be divided into several blocks automatically according to their statistical preferences, most of the existing multiblock modeling strategies tends to build local monitoring models individually, where the relevance among different blocks is ignored, and this leaves a room for enhancing process monitoring performance. Inspired by the recognition of this lack, a modified multiblock principal component analysis (MBPCA) algorithm is proposed for extracting block scores with respect to both specificity in each block and relevance among different blocks. Based on this sort of modeling strategy, a novel decentralized process monitoring is formulated by incorporating a PCA-based process decomposition strategy for block division, Bayesian inference to achieve decision fusion of fault detection, and reconstruction-based contribution plots for fault diagnosis. The superiority and validity of the proposed method is finally demonstrated through comparison studies on two simulated examples.

EMD-based online Filtering of Process Data

In chemical industries, measurements corrupted by noise or outliers may affect operators’ recognition of the current situation and lead them to make inappropriate control decisions. Data quality is a critical factor for process monitoring and fault diagnosis. A robust online filtering method (OLREMD11OLREMD: robust EMD-based online filtering of process data.) is proposed to implement online process data rectification with Empirical Mode Decomposition (EMD) as the basic algorithm. Tests with synthetic data show that OLREMD performs robustly with a lower sensitivity to parameters and improved performance on elimination of both noise and outliers. When applied to an industrial de-ethanizing column, OLREMD is shown to enhance the process monitoring performance.

Virtual enterprise collaborative processes monitoring through a project business approach

In order to design a system for managing performance in collaborative project-based enterprises, it is necessary to undertake real-time monitoring of its business processes and activities. This paper presents a systematic approach to project business process monitoring (BPM) and identifies the key aspects of virtual enterprise (VE) process evaluation. A framework for VE BPM is presented with special emphasis in linking interest groups to the development of their targets, information and knowledge sharing. The proposed model defines the exclusive performance metrics, which are needed during BPM. This interdisciplinary study examines BPM through peer-to-peer information exchange in the VE domain, which is currently a research gap. The fundamental metrics to define business process performance monitoring are elaborated in the research reported in this paper. The identified performance metrics can be used to measure the overall performances for both a project business and VE. A reference architecture is also highlighted with a case example with the objective to measure the performance in a project business VE. An overall definition of collaborative BPM and performance management systems is also presented accordingly. Future research directions are identified regarding the nature of collaboration in a project business VE and the characteristics of performance indicators to support it.

Autoregressive Dynamic Latent Variable Models for Process Monitoring

In most industrial processes, both autocorrelations and cross correlations in the data need to be extracted for the purpose of process monitoring and diagnosis. However, traditional dynamic modeling methods focus on the dynamic relationship while the cross correlations are at best implicit. In this brief, a new autoregressive dynamic latent variable model is proposed to capture both dynamic and static relationships simultaneously. The proposed method is a rather general dynamic model which can improve the performance of modeling and process monitoring. The Kalman filter and smoother are employed for inference while model parameters are estimated with an expectation-maximization algorithm. The corresponding fault detection method is also developed and a numerical example and the Tennessee Eastman benchmark process are used to evaluate the performance of the proposed model.

An Efficient Quality-Related Fault Diagnosis Method for Real-Time Multimode Industrial Process

Focusing on quality-related complex industrial process performance monitoring, a novel multimode process monitoring method is proposed in this paper. Firstly, principal component space clustering is implemented under the guidance of quality variables. Through extraction of model tags, clustering information of original training data can be acquired. Secondly, according to multimode characteristics of process data, the monitoring model integrated Gaussian mixture model with total projection to latent structures is effective after building the covariance description form. The multimode total projection to latent structures (MTPLS) model is the foundation of problem solving about quality-related monitoring for multimode processes. Then, a comprehensive statistics index is defined which is based on the posterior probability of the monitored samples belonging to each Gaussian component in the Bayesian theory. After that, a combined index is constructed for process monitoring. Finally, motivated by the application of traditional contribution plot in fault diagnosis, a gradient contribution rate is applied for analyzing the variation of variable contribution rate along samples. Our method can ensure the implementation of online fault monitoring and diagnosis for multimode processes. Performances of the whole proposed scheme are verified in a real industrial, hot strip mill process (HSMP) compared with some existing methods.

Using the expected detection delay to assess the performance of different multivariate statistical process monitoring methods for multiplicative and drift faults

Using the expected detection delay (EDD) index to measure the performance of multivariate statistical process monitoring (MSPM) methods for constant additive faults have been recently developed. This paper, based on a statistical investigation of the T2- and Q-test statistics, extends the EDD index to the multiplicative and drift fault cases. As well, it is used to assess the performance of common MSPM methods that adopt these two test statistics. Based on how to use the measurement space, these methods can be divided into two groups, those which consider the complete measurement space, for example, principal component analysis-based methods, and those which only consider some subspace that reflects changes in key performance indicators, such as partial least squares-based methods. Furthermore, a generic form for them to use T2- and Q-test statistics are given. With the extended EDD index, the performance of these methods to detect drift and multiplicative faults is assessed using both numerical simulations and the Tennessee Eastman process.

Nonlinear and robust statistical process monitoring based on variant autoencoders

Autoencoders (AEs) are an effective means for nonlinear feature extraction and dimension reduction. Variant autoencoders are an improvement over traditional AEs in terms of robustness. This paper proposes a novel nonlinear and robust process-monitoring approach based on variant autoencoders (variant AEs), which include denoising autoencoders (DAE) and contractive autoencoders (CAE). The CAE and DAE are powerful for extracting robust and nonlinear feature representations or manifold structures underlying data from industrial processes. Next, an online monitoring model is built through constructing new test statistic H2 based on the robust feature representations. The control limits are determined by kernel density estimation. The proposed method was applied to the Tennessee Eastman process (TE process) to evaluate its monitoring performance, and it demonstrated outstanding process-monitoring performance through the experimental results, especially for the barely detectable faults, such as 3, 5, 9, 10, 11, 15, 19, 20 and 21. Variant AEs monitoring provides a simple and very effective process-monitoring method for industrial processes.

복잡한 공정 측정 데이터의 실시간 분석을 통한 공정 감시

On-line process monitoring schemes are designed to give early warnings of process faults. In the artificial intelligence and machine learning fields, reliable approaches have been utilized, such as kernel-based nonlinear techniques. This work presents a kernel-based empirical monitoring scheme with a small sample problem. The measurement data of normal operations are easy to collect, whereas special events or faults data are difficult to collect. In such situations, noise filtering techniques can be helpful in enhancing the process monitoring performance. This can be achieved by the preprocessing of raw process data and eliminating unwanted variations of data. In this work, the performance of several monitoring schemes was demonstrated using three-dimensional batch process data. The results showed that the monitoring performance was improved significantly in terms of the detection success rate.

Process monitoring using statistical stability metrics: Applications to biopharmaceutical processes

ABSTRACTIndustrial manufacturing commonly employs control charts to monitor process performance, where special-cause variation is identified with runs rules. While the sensitivity of a control chart to detect non-random variation is increased as the number of rules increases, the false positive rate also increases. Accordingly, there is no way to discern a false signal from a true shift by simply observing runs rule violations on a control chart alone. In this article, two stability metrics are utilized to identify underlying variation as “common cause” or “special cause,” when Nelson rule violations are observed in an XmR control chart in biopharmaceutical processes.

Real-time business activity monitoring and analysis of process performance on big-data domains

Real-time access to business performance information is critical for corporations to run a competitive business and respond to a continuously changing business environment with ever-higher levels of competition. The timely analysis and monitoring of business processes are essential to identify non-compliant situations and react immediately to those inconsistencies in order to respond quickly to competitors. In this regard, the integration of business intelligence (BI) systems with Process Aware Information Systems (PAIS) can become a key tool for business users in decision making. However, current BI systems are not suitable for optimising and improving end-to-end processes since these are normally business domain specific and are not sufficiently process-aware to support the needs of process improvement type activities. In addition, highly transactional business environments may produce vast amounts of event data that cannot be efficiently managed by the use of traditional storage systems which are not designed to manage vast amounts of event data. We introduce a cloud-based architecture that leverages big-data technology to support performance analysis on any business domain, in a timely manner and regardless of the underlying concerns of the operational systems. Likewise, we demonstrate the ability of the solution to provide real-time business activity monitoring on big-data environments with low hardware costs.

Improved performance of process monitoring based on selection of key principal components

Conventional principal component analysis (PCA) can obtain low-dimensional representations of original data space, but the selection of principal components (PCs) based on variance is subjective, which may lead to information loss and poor monitoring performance. To address dimension reduction and information preservation simultaneously, this paper proposes a novel PC selection scheme named full variable expression. On the basis of the proposed relevance of variables with each principal component, key principal components can be determined. All the key principal components serve as a low-dimensional representation of the entire original variables, preserving the information of original data space without information loss. A squared Mahalanobis distance, which is introduced as the monitoring statistic, is calculated directly in the key principal component space for fault detection. To test the modeling and monitoring performance of the proposed method, a numerical example and the Tennessee Eastman benchmark are used.

A non-contact method for part-based process performance monitoring in end milling operations

Surface response to excitation (SuRE) method was originally developed for structural health monitoring (SHM) applications. SuRE was used to evaluate the performance of completed milling operations. The method generates surface waves on the plate and studies the spectrum changes at selected points to detect defects and change of compressive forces. In this study, the length, depth, and width of a slot were changed step by step. The surface of the aluminum plate was excited in the 20–400 kHz range with a piezoelectric element. A laser scanning vibrometer was used to monitor the vibrations at the predetermined grid points after the dimensions of the slot were changed methodically. The frequency spectrums of measured vibrations were calculated by using the Fast Fourier Transformation (FFT). The sums of the squares of the differences (SSD) of the spectrums were calculated to evaluate the change of the spectrums. The SuRE method was able to determine if the dimensions were changed in each case at all the selected points. The scanning laser vibrometer is not feasible to be used at the shop floor. However, the study demonstrated that a piezoelectric element attached to any of the grid points would be able to evaluate the completed machining process.