Manufacturing Process Control Research Articles

Additive Manufacturing in Australian Small to Medium Enterprises: Vat Polymerisation Techniques, Case Study and Pathways to Industry 4.0 Competitiveness

The advent of additive manufacturing (AM) in Australian small and medium-sized enterprises offers the direct benefits of time-saving and labour cost-effectiveness for Australian manufacturing to be highly competitive in global markets. Australian local businesses can tailor their products to a diverse range of customers with a quicker lead time on the sophisticated design and development of products under good quality control in the whole advanced manufacturing process. This review outlines typical AM techniques used in Australian manufacturing, which consist of vat polymerisation (VP), environmentally friendly AM, and multi-material AM. In particular, a practical case study was also highlighted in the Australian jewellery industry to demonstrate how manufacturing style is integrated into their manufacturing processes for the purpose of reducing lead time and cost. Finally, major obstacles encountered in AM and future prospects were also addressed to be well positioned as a key player in the revolutionised Industry 4.0.

Modeling Battery Formation: Boosted SEI Growth, Multi-Species Reactions, and Irreversible Expansion

This work proposes a semi-empirical model for the SEI growth process during the early stages of lithium-ion battery formation cycling and aging. By combining a full-cell model which tracks half-cell equilibrium potentials, a zero-dimensional model of SEI growth kinetics, and a semi-empirical description of cell thickness expansion, the resulting model replicated experimental trends measured on a 2.5 Ah pouch cell, including the calculated first-cycle efficiency, measured cell thickness changes, and electrolyte reduction peaks during the first charge dQ/dV signal. This work also introduces an SEI growth boosting formalism that enables a unified description of SEI growth during both cycling and aging. This feature can enable future applications for modeling path-dependent aging over a cell’s life. The model further provides a homogenized representation of multiple SEI reactions enabling the study of both solvent and additive consumption during formation. This work bridges the gap between electrochemical descriptions of SEI growth and applications toward improving industrial battery manufacturing process control where battery formation is an essential but time-consuming final step. We envision that the formation model can be used to predict the impact of formation protocols and electrolyte systems on SEI passivation and resulting battery lifetime.

Quality Control Optimization for Minimizing Security Risks Associated with Mesenchymal Stromal Cell-Based Product Development.

Mesenchymal stromal cells (MSCs) have been considered a therapeutic strategy in regenerative medicine because of their regenerative and immunomodulatory properties. The translation of MSC-based products has some challenges, such as regulatory and scientific issues. Quality control should be standardized and optimized to guarantee the reproducibility, safety, and efficacy of MSC-based products to be administered to patients. The aim of this study was to develop MSC-based products for use in clinical practice. Quality control assays include cell characterization, cell viability, immunogenicity, and cell differentiation; safety tests such as procoagulant tissue factor (TF), microbiological, mycoplasma, endotoxin, genomic stability, and tumorigenicity tests; and potency tests. The results confirm that the cells express MSC markers; an average cell viability of 96.9%; a low expression of HLA-DR and costimulatory molecules; differentiation potential; a high expression of TF/CD142; an absence of pathogenic microorganisms; negative endotoxins; an absence of chromosomal abnormalities; an absence of genotoxicity and tumorigenicity; and T-lymphocyte proliferation inhibition potential. This study shows the relevance of standardizing the manufacturing process and quality controls to reduce variability due to the heterogeneity between donors. The results might also be useful for the implementation and optimization of new analytical techniques and automated methods to improve safety, which are the major concerns related to MSC-based therapy.

Deep reinforcement learning optimized double exponentially weighted moving average controller for chemical mechanical polishing processes

This study investigates a deep reinforcement learning (DRL)-assisted double exponentially weighted moving average (dEWMA) controller for run-to-run (RtR) control in the semiconductor manufacturing process. We focus on implementing parameter adaption of dEWMA controllers to achieve disturbance compensation and target tracking. Owing to the powerful adaptive decision-making capability of the DRL, the weight adjustment of dEWMA controller is formulated as a Markov decision process. Specifically, the DRL behaves as an assisted controller to derive appropriate weights that facilitate dEWMA to perform highly accurate disturbance estimation, whereas the standard dEWMA works as a baseline controller to provide suitable recipes for the manufacturing process. Consequently, a composite control strategy integrating DRL and dEWMA is developed. In addition, a twin-delayed deep deterministic policy gradient algorithm is employed to adjust the weights of dEWMA online. The effectiveness of the proposed scheme is validated in a chemical mechanical polishing process. Several disturbance rejection scenarios verify the benefits of the suggested approach.

Automated calibration and in-line measurement of product quality during therapeutic monoclonal antibody purification using Raman spectroscopy.

Current manufacturing and development processes for therapeutic monoclonal antibodies demand increasing volumes of analytical testing for both real-time process controls and high-throughput process development. The feasibility of using Raman spectroscopy as an in-line product quality measuring tool has been recently demonstrated and promises to relieve this analytical bottleneck. Here, we resolve time-consuming calibration process that requires fractionation and preparative experiments covering variations of product quality attributes (PQAs) by engineering an automation system capable of collecting Raman spectra on the order of hundreds of calibration points from two to three stock seed solutions differing in protein concentration and aggregate level using controlled mixing. We used this automated system to calibrate multi-PQA models that accurately measured product concentration and aggregation every 9.3 s using an in-line flow-cell. We demonstrate the application of a nonlinear calibration model for monitoring product quality in real-time during a biopharmaceutical purification process intended for clinical and commercial manufacturing. These results demonstrate potential feasibility to implement quality monitoring during GGMP manufacturing as well as to increase chemistry, manufacturing, and controls understanding during process development, ultimately leading to more robust and controlled manufacturing processes.

Multi-physics data registration for the improvement of Additive Manufacturing process control

The work presented in this paper is to tend towards a goal of developing a methodology and associated software module to establish cross-correlation between real time process monitoring data and characterization data. To implement this methodology, a number of steel specimen were built following a design of experiment approach. During the manufacturing, all monitoring data was recorded in log files with a precise timestamp, the data collected being then temporal signals. Subsequently, these parts were inspected by high resolution X-ray tomography, and the defects present in the part were characterized, resulting in 3D volumetric dataset. The software module developed on CIVA platform then enables the registration of these multi-physics data.

Common chemistry, manufacturing, and control deficiencies in abbreviated new drug applications assessed by the US Food and drug administration: Hurdle to access cost-effective medicines

To market a generic product in the United States, it must be registered in Common Technical Document (CTD) format with the US Food and Drug Administration. The Generic Drug User Fee Act went into force in 2012, to expedite the timely review of Abbreviated New Drug Applications (ANDA) by communicating potential defects in the application to the applicant through deficiency letters at different time intervals during the review cycle. This often delays product approval since these deficiencies must be resolved before the product can be approved. In the present study, a study was performed to analyze the recurrent queries for ANDA applications in the CTD quality module from 2013 to 2020, and the probable corrective and preventive action to be taken was drafted. The most frequently occurring queries were observed in the sections titled “Description of manufacturing process and process controls”, “Controls of critical steps and intermediates”, “Specifications (Control of drug product)”, and “Stability data”.

SSMSPC: self-supervised multivariate statistical in-process control in discrete manufacturing processes

Self-supervised learning has demonstrated state-of-the-art performance on various anomaly detection tasks. Learning effective representations by solving a supervised pretext task with pseudo-labels generated from unlabeled data provides a promising concept for industrial downstream tasks such as process monitoring. In this paper, we present SSMSPC a novel approach for multivariate statistical in-process control (MSPC) based on self-supervised learning. Our motivation for SSMSPC is to leverage the potential of unsupervised representation learning by incorporating self-supervised learning into the general statistical process control (SPC) framework to develop a holistic approach for the detection and localization of anomalous process behavior in discrete manufacturing processes. We propose a pretext task called Location + Transformation prediction, where the objective is to classify both, the type and the location of a randomly applied augmentation on a given time series input. In the downstream task, we follow the one-class classification setting and apply the Hotelling’s T2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$T^2$$\\end{document} statistic on the learned representations. We further propose an extension to the control chart view that combines metadata with the learned representations to visualize the anomalous time steps in the process data which supports a machine operator in the root cause analysis. We evaluate the effectiveness of SSMSPC with two real-world CNC-milling datasets and show that it outperforms state-of-the-art anomaly detection approaches, achieving 100%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$100\\%$$\\end{document} and 99.6%\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$99.6\\%$$\\end{document} AUROC, respectively. Lastly, we deploy SSMSPC at a CNC-milling machine to demonstrate its practical applicability when used as a process monitoring tool in a running process.

Distributional reinforcement learning for run-to-run control in semiconductor manufacturing processes

Distributional reinforcement learning for run-to-run control in semiconductor manufacturing processes

Real-time prediction and adaptive adjustment of continuous casting based on deep learning

Digitalisation of metallurgical manufacturing, especially technological continuous casting using numerical models of heat and mass transfer and subsequent solidification has been developed to achieve high manufacturing efficiency with minimum defects and hence low scrappage. It is still challenging to perform adaptive closed-loop process adjustment using high-fidelity computation in real-time. To address this challenge, surrogate models are a good option to replace the high-fidelity model, with acceptable accuracy and less computational time and cost. Based on deep learning technology, here we developed a real-time prediction (ReP) model to predict the three-dimensional (3D) temperature field distribution in continuous casting on millisecond timescale, with mean absolute error (MAE) of 4.19 K and mean absolute percent error (MAPE) of 0.49% on test data. Moreover, by combining the ReP model with machine learning technology—Bayesian optimisation, we realised the rapid decision-making intelligent adaptation of the operating parameters for continuous casting with high predictive capability. This innovative and reliable method has a great potential in the intelligent control of the metallurgical manufacturing process.

Implementation of cloud based IoT technology in manufacturing industry for smart control of manufacturing process

Implementation of cloud based IoT technology in manufacturing industry for smart control of manufacturing process

Rapid Vector-Based Peak Fitting and Resolution Enhancement for Correlation Analyses of Raman Hyperspectra.

Spectroscopic peak parameters are important since they provide information about the analyte under study. Besides obtaining these parameters, peak fitting also resolves overlapped peaks. Thus, the obtained parameters should permit the construction of a higher-resolution version of the original spectrum. However, peak fitting is not an easy task due to computational reasons and because the true nature of the analyte is often unknown. These difficulties are major impediments when large hyperspectral data sets need to be processed rapidly, such as for manufacturing process control. We have developed a novel and relatively fast two-part algorithm to perform peak fitting and resolution enhancement on such data sets. In the first part of the algorithm, estimates of the total number of bands and their parameters were obtained from a representative spectrum in the data set, using a combination of techniques. Starting with these parameter estimates, all the spectra were then iteratively and rapidly fitted with Gaussian bands, exploiting intrinsic features of the Gaussian distribution with vector operations. The best fits for each spectrum were retained. By reducing the obtained bandwidths and commensurately increasing their amplitudes, high-resolution spectra were constructed that greatly improved correlation-based analyses. We tested the performance of the algorithm on synthetic spectra to confirm that this method could recover the ground truth correlations between highly overlapped peaks. To assess effective peak resolution, the method was applied to low-resolution spectra of glucose and compared to results from high-resolution spectra. We then processed a larger spectral data set from mammalian cells, fixed with methanol or air drying, to demonstrate the resolution enhancement of the algorithm on complex spectra and the effects of resolution-enhanced spectra on two-dimensional correlation spectroscopy and principal component analyses. The results indicated that the algorithm would allow users to obtain high-resolution spectra relatively fast and permit the recovery of important aspects of the data's intrinsic correlation structure.

Stability analysis using mixed models: A critique of tolerance interval methods and a probabilistic solution.

Recently, tolerance interval approaches to the calculation of a shelf life of a drug product have been proposed in the literature. These address the belief that shelf life should be related to control of a certain proportion of batches being out of specification. We question the appropriateness of the tolerance interval approach. Our concerns relate to the computational challenges and practical interpretations of the method. We provide an alternative Bayesian approach, which directly controls the desired proportion of batches falling out of specification assuming a controlled manufacturing process. The approach has an intuitive interpretation and posterior distributions are straightforward to compute. If prior information on the fixed and random parameters is available, a Bayesian approach can provide additional benefits both to the company and the consumer. It also avoids many of the computational challenges with the tolerance interval methodology.

Digital Twin Implementation for Machining Center Based on ISO 23247 Standard

In the context of the Industry 4.0, Digital Twins are essential tools for monitoring and controlling manufacturing processes. In this sense, this work presents the development an implementation of a Digital Twins architecture for machining using a CNC Haas Mini Mill machine. The proposed architecture is based on the current Digital Twins framework for manufacturing normalized by ISO 23247. MQTT and MTConnect protocols are used to transfer machine status and process data to cloud platforms for realizing storing, monitoring, visualization, and simulation activities. This approach was validated in a real CNC machine and it is feasible to be applied to other machines and manufacturing processes.

Summary of Research on Digital Twin Workshop

Global overcapacity, resulting in increasingly fierce competition between enterprises, improve the production efficiency of the manufacturing industry is the key problem. Workshop is an important part of manufacturing industry. In order to improve the intelligent level of workshop and realize the digitization of production process, the current focus is how to use digital twin technology. Although the traditional virtual shop and digital shop design can realize intelligent shop, there are some problems such as non-real-time interaction and low data utilization. Digital twinning technology can effectively improve the transparency of workshop production process and optimize the production process. Establishing digital twin workshops and realizing the interconnection and further integration of workshop information and physical space will become the development trend of workshops and the only way to realize intelligent production and control of workshops. This study describes the definition of digital twin shop, system architecture and supporting technology, and summarizes the applications of digital twin shop in six aspects: shop equipment fault prediction and maintenance, shop production scheduling optimization, production process planning, intelligent control of shop equipment, shop energy consumption analysis and optimization, and real-time control of shop manufacturing process. This will provide reference for later digital twinning workshop researchers.

Modular supervisory control for multi-floor manufacturing processes

Due to space availability limitations and high land costs, there is an increasing development of multi-floor manufacturing (MFM) systems in urban and industrial areas. The problem of coordination in a multi-floor manufacturing process, in the Ramadge Wonham framework, is introduced. The manufacturing chain of each floor and the elevator system are modeled in the form of finite deterministic automata. The models of the multi-floor manufacturing process are parametric with respect to the number of floors and the number of manufacturing machines on each floor. The coordination desired performance is formulated in the form of desired regular languages in analytic forms. The languages are realized by appropriate supervisors in the form of finite deterministic automata. The models of the supervisors are also parametric with respect to the number of floors and the number of manufacturing machines on each floor. The total control of the coordination of the multi-floor manufacturing process is accomplished via a modular supervisory control architecture. The complexity of the supervisors as well as the complexity of the total modular supervisory architecture are determined in analytic forms with respect to the number of floors and the number of manufacturing machines on each floor. The special case of a two floor manufacturing process is presented as an illustrative example.

Data-driven methods for stress field predictions in random heterogeneous materials

Predicting full-field stress responses is of fundamental importance to assessing materials failure and has various engineering applications in design optimization, manufacturing process control, and structural health monitoring. This article develops and evaluates different data-driven methods for efficient and accurate predictions of full stress fields in random heterogeneous materials. The first approach integrates model order reduction of proper orthogonal decomposition (POD) with classical machine learning techniques (K-nearest neighbors, random forest, and artificial neural networks) to predict full-field responses based on POD-reduced coefficients. However, this strategy shows limitations in predicting full stress fields, especially for heterogeneous material inclusions of small size or being close to the domain boundary. After that, two computer vision-based deep learning approaches were developed for full-field predictions. The first one uses a Resnet-based Convolutional Neural Network (CNN), and the second is based on a modified conditional Generative Adversarial Network (cGAN). Two representative example problems were studied: a random heterogeneous material inclusion or a void varying in size and location. In contrast to POD-based classical machine learning, almost invisible differences were found between the entire stress fields in finite element simulations and computer vision-based deep learning (CNN/cGAN) predictions, with significantly reduced mean squared error (MSE) and correlation values (R2) mostly above 0.99. On the other hand, the proposed cGAN provides more accurate predictions than CNN with fewer epochs.

The nature of a low angle grain boundary in a Si bi-crystal with added Fe impurities

Silicon solar cell performance can be severely degraded when low-angle grain boundaries and impurities are present in the material. These two factors often come hand in hand, and it is imperative to understand the underlying fundamental physics to make progress in the field of solar energy conversion. In this study, three different types of Lomer dislocations in a 5° Si grain boundary were identified by atomic resolution HAADF-STEM. DFT calculations and HAADF-STEM revealed that these grain boundaries are not perfectly flat and must exhibit additional shifts along <110>Si to account for the misorientation angle. DFT simulations of the Lomer type I dislocation core were in excellent agreement with the atomic structure characterised by HAADF-STEM.The nature of the segregation of impurity elements Fe and C at the grain boundary is established by atom probe tomography, DFT calculations and HAADF-STEM studies. DFT calculations predict that Fe can segregate to all interstitial sites surrounding the dislocation core except for one, while C would prefer three out of ten possible substitutional sites. These results are corroborated by atom probe tomography. Additional impurity elements were also discovered at the grain boundaries, illustrating the importance of a controlled manufacturing process and crucible coating.

Transformer-enabled generative adversarial imputation network with selective generation (SGT-GAIN) for missing region imputation

Although data have been extensively leveraged for process monitoring and control in advanced manufacturing, it still suffers from the connection issues among sensors, machines, and computers, which may lead to significant data loss, i.e., missing region in the collected data, in the application of data-driven monitoring. To address the missing region issues, one popular way is to perform missing data imputation. With the advances of machine learning, many approaches have been developed for the missing data imputation, such as the popular Generative Adversarial Imputation Network (GAIN), which is based on the Generative Adversarial Network (GAN). However, the inherent shortcomings of generative adversarial architecture may still lead to unstable training. More importantly, the collected online sensor data in manufacturing are in sequential order whereas GAIN considered the input data independently. Hence, to address these two limitations, this work proposes a novel approach termed transformer-enabled GAIN with selective generation (SGT-GAIN). The contributions of the proposed SGT-GAIN consist of three aspects: (i) the architecture for transformer-enabled generation is developed to capture the sequential information among the data; (ii) a selective multi-generation framework is proposed to further reduce the imputation bias; and (iii) an ensemble learning framework is applied to enhance the imputation robustness. Both the numerical simulation study and a real-world case study in additive manufacturing demonstrated the effectiveness of the proposed SGT-GAIN.

Numerical predictions and experimental validation on the effect of material properties in filament material extrusion

The numerical prediction of dimensional accuracy and mechanical performance of 3D-printed structures can eliminate or reduce the need to perform trial-and-error experimentation to achieve the desired part quality quickly and at a reduced cost. Therefore, this study focuses on experimental and numerical investigations of the effect of material properties on dimensional control of the material extrusion additive manufacturing process. A thermomechanical model is utilized, validated, and used to predict the part deflections, warpages, and residual stresses in 3D-printed components for two different materials (i.e., polyamide-6 (PA6) and acrylonitrile butadiene styrene (ABS)). In addition, numerical simulations for mechanical testing were performed and validated through experimental investigations. The numerical model considered the effect of material on final product quality and was in close agreement with experimental results. It was concluded that the materials with a lower thermal expansion coefficient (CTE) would provide better dimensional accuracy of 3D-printed parts. In addition, the lower thermal conductivity and temperature variations within the 3D-printed part will further produce precise parts. The controlled ambient environment using a closed chamber can also assist in maintaining desired thermal gradient and avoid non-uniform cooling during the process. Finally, the numerical model accurately predicted the mechanical behavior of 3D-printed samples as validated via experimentation.