Process Control Strategies Research Articles

An artificial neural network approach for out‐of‐control stream identification in multiple stream processes

AbstractA multiple stream process (MSP) is a process at a point in time that generates several streams of output with quality variables and specifications that are identical in all streams. Statistical process control (SPC) schemes for MSPs are designed to issue an out‐of‐control (OC) signal when a possible change occurs from the in‐control (IC) state of the process. However, the SPM literature for MSPs lacks contributions regarding post‐hoc identification of the stream or group of streams responsible for an OC signal. In this article, we propose an artificial neural network (NN) specifically trained for this purpose and, through a Monte Carlo simulation, show its superiority in correctly identifying OC streams, since an OC signal has been correctly issued. The research is motivated by the need for a post hoc diagnosis of the heating, ventilation, and air conditioning (HVAC) systems installed on modern train coaches, generating multiple streams of quality variables of interest. A case study in this area based on the HVAC data (openly available online at https://github.com/unina‐sfere/NN4OCMSP) illustrates the practical applicability of the proposed approach, which is implemented in the Python package NN4OCMSP and published on the PyPI software repository.

Industrial data-driven machine learning soft sensing for optimal operation of etching tools

Smart Manufacturing, or Industry 4.0, has gained significant attention in recent decades with the integration of Internet of Things (IoT) and Information Technologies (IT). As modern production methods continue to increase in complexity, there is a greater need to consider what variables can be physically measured. This advancement necessitates the use of physical sensors to comprehensively and directly gather measurable data on industrial processes; specifically, these sensors gather data that can be recontextualized into new process information. For example, artificial intelligence (AI) machine learning-based soft sensors can increase operational productivity and machine tool performance while still ensuring that critical product specifications are met. One industry that has a high volume of labor-intensive, time-consuming, and expensive processes is the semiconductor industry. AI machine learning methods can meet these challenges by taking in operational data and extracting process-specific information needed to meet the high product specifications of the industry. However, a key challenge is the availability of high quality data that covers the full operating range, including the day-to-day variance. This paper examines the applicability of soft sensing methods to the operational data of five industrial etching machines. Data is collected from readily accessible and cost-effective physical sensors installed on the tools that manage and control the operating conditions of the tool. The operational data are then used in an intelligent data aggregation approach that increases the scope and robustness for soft sensors in general by creating larger training datasets comprised of high value data with greater operational ranges and process variation. The generalized soft sensor can then be fine-tuned and validated for a particular machine. In this paper, we test the effects of data aggregation for high performing Feedforward Neural Network (FNN) models that are constructed in two ways: first as a classifier to estimate product PASS/FAIL outcomes and second as a regressor to quantitatively estimate oxide thickness. For PASS/FAIL classification, a data aggregation method is developed to enhance model predictive performance with larger training datasets. A statistical analysis method involving point-biserial correlation and the Mean Absolute Error (MAE) difference score is introduced to select the optimal candidate datasets for aggregation, further improving the effectiveness of data aggregation. For large datasets with high quality data that enable model training for more complex tasks, regression models that predict the oxide thickness of the product are also developed. Two types of models with different loss functions are tested to compare the effects of the Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE) loss functions on model performance. Both the classification and regression models can be applied in industrial settings as they provide additional information regarding the process outcome. Individually, these models can reduce the number of metrology steps in semiconductor factories, and when developed further, can empower the development of advanced process control strategies.

Beyond viability: Advancing CHO cell culture process strategies to modulate host cell protein levels

Chinese Hamster Ovary (CHO) cells are widely utilized in bioprocessing industry for monoclonal antibody (mAb) production. In many instances, challenges persist in achieving sufficient clearance of Host Cell Proteins (HCPs) in the final drug substance. While purification strategies usually offer substantial HCP clearance, certain "problematic" HCPs, particularly lipases, continue to pose significant challenges. This study investigates the accumulation of various "problematic" HCPs in CHO cell culture using transcriptomics, revealing correations between cell culture parameters and HCP level. Contrary to conventional assumptions, viability alone does not reliably predict HCP levels, with factors such as clone selection, host cell line choice, media and feed compositions significantly influencing HCP accumulation. Leveraging transcriptomics-based approaches, we demonstrate the potential of upstream process control strategies to mitigate HCP presence and improve biologic product quality. Our findings underscore the importance of considering diverse cell culture parameters in bioprocess optimization to ensure product stability and quality. While promising, further research is needed to elucidate the mechanisms underlying HCP release and propagation through downstream processing stages, emphasizing the necessity of a comprehensive integrated approach to HCP control in biologics production.

Multivariate SPC via sequential multiblock-PLS

The sequential multi-block partial least squares (SMB-PLS) is proposed for implementing a multivariate statistical process control scheme. This is of interest when the system is composed of several blocks following a sequential order and presenting correlated information, for instance, a raw material properties block followed by a process variables block that is manipulated according to raw material properties. The SMB-PLS uses orthogonalization to separate correlated information between blocks from orthogonal variations. This allows monitoring the system in different stages considering only the remaining orthogonal part in each block. Thus, the SMB-PLS increases the interpretability and process understanding in the model building (Phase I), since it provides a deep insight about the nature of the system variations. Besides, it prevents any special cause from propagating to subsequent blocks enabling their use in the model exploitation (Phase II). The methodology is applied to a real case study from a food manufacturing process.

Scaling laws for nanoparticles – Online shape heterogeneity analysis by size exclusion chromatography coupled with multi-angle light scattering

Nanoparticle-based drug delivery systems are rising technologies to access challenging therapeutic targets. Following commercial success of lipid-based nanoparticles (LBNP), accruing understandings of nanoparticle structures and critical quality attributes through advanced analytics are beneficial to future clinical development and generalization of this delivery platform. The morphological attributes of nanoparticles, such as shape, can affect uptake, cell-interaction, drug release, circulation, and flow. Gaining an understanding of these structure-activity relationships in early-stage formulation development is important because mix morphologies can affect quality and potency but often exist before process control strategies are fully implemented. In this study, we used shape heterogeneous nanoparticle mixtures, containing various populations of liposomes and lipodisks, as a model system and developed an online semi-quantitative method to characterize the nanoparticle shape heterogeneity by size exclusion chromatography (SEC) coupled with multi-angle light scattering (MALS). The liposomes and lipodisks were separated in SEC when their sizes were ∼3 fold different. When the particles of different shapes were in similar sizes, size-based separation was not always feasible. Instead, light scattering data distinguished liposomes and lipodisks by the scaling law linking radius of gyration and molecular weight of the nanoparticles, enabling morphological identification. A semi-quantitative model was built based on the exponential correlation between the scaling law exponents and the ratios of liposomes and lipodisks. The model was applied to test 6 random formulations made with different compositions and manufacturing processes, and the predicted liposome percentage for 5 formulations was within 25 % absolute difference from the percentage determined by cryogenic electron microscopy (cryo-EM). We envision this method being routinely used to characterize liposome and lipodisk shape heterogeneity during formulation screening as well as on stability studies. Potentially, the method can be converted to in-process control method and extended to other categories of nanoparticles beyond liposomes.

Optimization of Quality Process Control and Preventive Maintenance Strategy: A Case Study

Optimization of Quality Process Control and Preventive Maintenance Strategy: A Case Study

Monitoring power consumption with automated process control strategy and adjustment considering price sensitivity and multi-source collaboration

In context of demand side management, the real-time pricing is an effective method to achieve power peak-clipping and valley-filling and maintain power balance between supply and demand. However, in reality, users’ power consumption fluctuation may lead to collapse of power systems and waste of resources. In this paper, based on the cosine function for demand response program, a novel two-stage monitoring and adjustment strategy is proposed to monitor power consumption, which considers the changes of users’ sensitivity to the electricity price and power generation with multi-source collaboration in detail. When the forecast power consumption exceeds the upper or lower bound, we design a price adjustment strategy with automated process control to better guide users to modify their electricity consumption behavior in the first stage, and a generation adjustment strategy with multi-source collaboration is offered to meet the adjusted consumption demand and low-carbon development in the second stage. Simulation results indicate that the proposed two-stage strategy not only has great performance of peak-clipping and valley-filling, but also is conductive to social economy and environment. The carbon emission decreases by 7.5%. The social welfare and supplier’s profit increase by 2.42% and 31.38% after monitoring and adjustment, respectively. Compared with the linear, quadratic and logarithmic strategies, the proposed two-stage strategy is better to encourage users to consume electricity appropriately.

Identifying the pollution risk pattern from industrial production to rural settlements and its countermeasures in China

Impacted by large-scale and rapid rural industrialization in the past few decades, China's rural settlements are confronted with the risk of heavy metal pollution stemming from industrial production, which might pose a significant threat to the rural habitat and the well-beings. This study devised a relative risk model for industrial heavy metal pollution to the rural settlements based on the source-pathway-receptor risk theory. Using this model, we assessed the risk magnitudes of heavy metal pollution from industrial production at a 10 km × 10 km grid scale and identified the characteristics of the risk pattern in China. Our finding reveals: (1) the relative risk values of wastewater, waste gas and total heavy metal pollution are notably concentrated within a confined spectrum, with only a small number of units are characterized by high-risk level; (2) Approximately 21.57 % of China's rural settlements contend with heavy metal pollution, with 4.17 %, 9.84 % and 7.55 % being subjected to high, medium and low risks, respectively; (3) The high-risk units mainly is concentrated in the developed areas such as Yangtze River Delta, Pearl River Delta, and the Beijing-Tianjin metropolitan area, also dispersed in the plain areas with high rural population density. Guided by these insights, this study puts forth regionally tailored prevention and control strategies, as well as distinct process prevention and control strategies.

Overview and Potential Solutions of Current and Future Challenges in CMP

This talk is dedicated to the special session in honor of Prof. Babu Shrink of critical dimensions and stringent technology specifications are posing several planarization challenges in technology development (TD) and high-volume manufacturing (HVM) phases of current and advanced technology nodes. In this talk, we will focus on process challenges related to achieving (a) minimum defects (b) high planarization efficiency with less than 1nm dishing (c) tighter WIW and WID uniformities (d) ultra-smooth post-CMP surface roughness in the order of sub-nanometer. In addition, identifying appropriate endpoint control techniques combined with advanced process control strategies to provide good process margin will also be discussed. Addressing the above challenges in a cost-effective manner is vital for HVM. Root cause analysis and probable working models will be discussed for high-risk process challenges. Based on our understanding, a jist shall be provided about challenges associated with future technology nodes in logic and memory industries, with focus on how (a) technical understanding, (b) change in CMP ecosystem, and (c) AI can help solving technology development and high-volume manufacturing problems.

MFRL-BI: Design of a model-free reinforcement learning process control scheme by using Bayesian inference

Design of the process control scheme is critical for quality assurance to reduce variations in manufacturing systems. Taking semiconductor manufacturing as an example, extensive literature focuses on control optimization based on certain process models (usually linear models), which are obtained by experiments before a manufacturing process starts. However, in real applications, pre-defined models may not be accurate, especially for a complex manufacturing system. To tackle model inaccuracy, we propose a Model-Free Reinforcement Learning (MFRL) approach to conduct experiments and optimize control simultaneously according to real-time data. Specifically, we design a novel MFRL control scheme by updating the distribution of disturbances using Bayesian inference to reduce their large variations during manufacturing processes. As a result, the proposed MFRL controller is demonstrated to perform well in a nonlinear chemical mechanical planarization process when the process model is unknown. Theoretical properties are guaranteed when disturbances are additive. The numerical studies also demonstrate the efficiency of our methodology.

Readers’ awareness in the use of intertextual strategies when writing from multiple texts

Metacognitive reading awareness, involving cognitive process control and reading strategies, is linked to better comprehension and performance, but its relationship with intertextual integration strategies and the quality of argumentative essays remains unexplored. This study aimed to investigate the role of metacognition in employing integration strategies when reading conflicting texts. 69 undergraduate students participated in an online reading-writing activity, where they wrote argumentative essays based on conflicting texts about red meat consumption. We examined the students' use of intertextual integration strategies (refutation, weighing, synthesizing) and assessed their metacognitive awareness through their reflections on these strategies. The quality of the argumentative essays served as a measure of multiple text comprehension. The results indicated a lack of metacognitive awareness regarding integration strategies, with students overestimating their ability to employ these strategies. However, they demonstrated better understanding of refutational strategies based on the examples provided in their essays. Interestingly, students who were aware of and utilized these strategies in their essays performed better in the multiple-text comprehension task.

Developing deep learning surrogate models for digital twins in mineral processing – A case study on data-driven multivariate multistep forecasting

The escalating demand for environmental and social sustainability underscores the critical need for large industries such as mining and metallurgy to function optimally. Achieving optimal operation of a mineral processing plant requires timely access to information and rapid implementation of corrective actions. Essential information, such as current and future production, plays a pivotal role in enhancing process efficiency by facilitating adjustments in key aspects, including feed, control and operating variables. This study delves into the application of deep learning algorithms to forecast the gold concentrate grade of a flotation process half an hour in advance, a valuable tool for short-term decision-making, e.g., with regard to optimising the operation and process control. Multivariate time series data from a real gold processing plant was acquired through the process instrumentation and control system and augmented with simulated data at 5-minute intervals over 236 days. Surrogate models based on LSTM, GRU and CNN architectures were developed and analysed for predicting the gold grade one to six steps ahead. Evaluation of the testing dataset indicates that the CNN exhibits superior performance, achieving an average error reduction of 36 %, 45 % and 35 % for the performance metrics RMSE, MAE and MAPE respectively compared to the baseline model. These findings reveal the high potential of deep learning surrogate methods for accurately forecasting process outputs, enabling effective data-driven process control and optimisation strategies.

Digital Twin for Continuous Production of Virus-like Particles toward Autonomous Operation.

Lentiviral vector and virus-like particle (VLP) manufacturing have been published in fed-batch upstream and batch downstream modes before. Batch downstream and continuous upstream in perfusion mode were reported as well. This study exemplifies development and validation steps for a digital twin combining a physical-chemical-based mechanistic model for all unit operations with a process analytical technology strategy in order to show the efforts and benefits of autonomous operation approaches for manufacturing scale. As the general models are available from various other biologic manufacturing studies, the main step is model calibration for the human embryo kidney cell-based VLPs with experimental quantitative validation within the Quality-by-Design (QbD) approach, including risk assessment to define design and control space. For continuous operation in perfusion mode, the main challenge is the efficient separation of large particle manifolds for VLPs and cells, including cell debris, which is of similar size. Here, innovative tangential flow filtration operations are needed to avoid fast blocking with low mechanical stress pumps. A twofold increase of productivity was achieved using simulation case studies. This increase is similar to improvements previously described for other entities like plasmid DNAs, monoclonal antibodies (mAbs), and single-chain fragments of variability (scFv) fragments. The advantages of applying a digital twin for an advanced process control strategy have proven additional productivity gains of 20% at 99.9% reliability.

Quantitative modelling of Plato and total flavonoids in Qingke wort at mashing and boiling stages based on FT-IR combined with deep learning and chemometrics

Craft beer brewers need to learn process control strategies from traditional industrial production to ensure the consistent quality of the finished product. In this study, FT-IR combined with deep learning was used for the first time to model and analyze the Plato degree and total flavonoid content of Qingke beer during the mashing and boiling stages and to compare the effectiveness with traditional chemometrics methods. Two deep learning neural networks were designed, the effect of variable input methods on the effectiveness of the models was discussed. The experimental results showed that the CARS-LSTM model had the best predictive performance, not only as the best quantitative model for Plato in the mashing (R2p = 0.9368) and boiling (R2p = 0.9398) phases but also as the best model for TFC in the boiling phase (R2p = 0.9154). This study demonstrates the great potential of deep learning and provides a new approach to quality control analysis in beer brewing.

A MATLAB-based simulator for the study of process control of fed-batch yeast fermentations

This paper presents a MATLAB-based simulator for the study of process control of fed-batch yeast fermentations that meets the educational needs of undergraduate and graduate students. Against the background of challenges in interdisciplinary education, the paper examines the evolving environment of simulation tools in the field of bioprocesses. It emphasizes the need for interdisciplinarity to prepare students for the complexities of the modern biotechnology industry. Built with accessibility and flexibility in mind, the simulator offers a modular structure with a graphical user interface (GUI) for novice users and direct access to MATLAB functions for advanced users. This design choice ensures ease of use for students with different programming backgrounds and allows for adaptability to alternative software environments. The introductory sections provide a detailed overview of the simulator development, including the mathematical models that govern biomass growth kinetics, mass transfer, and process quality indicators. Selected process control strategies such as rule-based and feedforward approaches are incorporated in the simulator to enhance the learning experience by allowing students to experiment with different scenarios. The paper concludes by emphasizing the simulator's adaptability, modularity, and user-friendly interface as a valuable asset in educating students about the complexities of bioprocess control.

Application of fed-batch strategy to fully eliminate the negative effect of lignocellulose-derived inhibitors in ABE fermentation

BackgroundInhibitors that are released from lignocellulose biomass during its treatment represent one of the major bottlenecks hindering its massive utilization in the biotechnological production of chemicals. This study demonstrates that negative effect of inhibitors can be mitigated by proper feeding strategy. Both, crude undetoxified lignocellulose hydrolysate and complex medium supplemented with corresponding inhibitors were tested in acetone–butanol–ethanol (ABE) fermentation using Clostridium beijerinckii NRRL B-598 as the producer strain.ResultsFirst, it was found that the sensitivity of C. beijerinckii to inhibitors varied with different growth stages, being the most significant during the early acidogenic phase and less pronounced during late acidogenesis and early solventogenesis. Thus, a fed-batch regime with three feeding schemes was tested for toxic hydrolysate (no growth in batch mode was observed). The best results were obtained when the feeding of an otherwise toxic hydrolysate was initiated close to the metabolic switch, resulting in stable and high ABE production. Complete utilization of glucose, and up to 88% of xylose, were obtained. The most abundant inhibitors present in the alkaline wheat straw hydrolysate were ferulic and coumaric acids; both phenolic acids were efficiently detoxified by the intrinsic metabolic activity of clostridia during the early stages of cultivation as well as during the feeding period, thus preventing their accumulation. Finally, the best feeding strategy was verified using a TYA culture medium supplemented with both inhibitors, resulting in 500% increase in butanol titer over control batch cultivation in which inhibitors were added prior to inoculation.ConclusionProperly timed sequential feeding effectively prevented acid-crash and enabled utilization of otherwise toxic substrate. This study unequivocally demonstrates that an appropriate biotechnological process control strategy can fully eliminate the negative effects of lignocellulose-derived inhibitors.Graphical

In‐mold rheology and automated process control for injection molding of recycled polypropylene

AbstractManufacturing plastic parts with secondary feedstocks has risen to the forefront of importance in recent years. However, the variation in molecular weight and rheology of secondary feedstock can lead to inconsistent part quality. This work evaluates the effectiveness of a novel closed‐loop adaptive process control system that adjusts nozzle pressure in response to in‐mold pressure data. Five different recycled polypropylene blends, with a broad distribution of flow properties, were evaluated to determine the effectiveness of the control system at reducing processing variation. The experimental results show that the process control strategy reduced the variation within the mold, as seen by in‐mold pressure curves and calculated in‐mold viscosity values. Additionally, the parameters that control the automated process adjustments were investigated, showing the importance of optimization. The analysis of the correlation between in‐mold rheology and mechanical properties showed a slight variation in the mechanical properties and parts weight with a coefficient of variation of under 5%. Overall, the results demonstrate the ability of pressure‐controlled molding and automated viscosity adjustment to reduce the variability when molding a secondary feedstock.Highlights Pressure‐controlled injection molding of recycled polypropylene. Automated closed‐loop adaptive process control methodology. Methodology resulted in a reduction in pressure variation during molding. Changes in mechanical properties and in‐mold viscosity were investigated. Results show the potential of pressure‐controlled molding at reducing variation.

KT-Biologics I (KTB1): A dynamic simulation model for continuous biologics manufacturing

The pharmaceutical industry's shift towards biological therapeutics has led to a transition from conventional batch production to continuous manufacturing. This change highlights the crucial need for effective process monitoring and control strategies to ensure consistent product quality and stability. Open-source benchmark simulation models have become essential tools for refining these processes, offering a platform for testing research hypotheses. This study uses the production of Lovastatin as a case study for continuous biopharmaceutical production. A comprehensive dynamic model covering upstream and downstream components provides an integrated perspective of the production process. The study introduces a basic control system emphasizing realistic sensor and actuator integration to enhance simulation accuracy. It assesses the benchmark through open-loop and closed-loop simulations, offering an in-depth analysis of the KTB1 model's dynamic response and functionality. KTB1 represents a model-driven decision support tool that enables the evaluation of monitoring strategies, process design, process optimization, and control for biomanufacturing.

Integrating artificial intelligence modeling and membrane technologies for advanced wastewater treatment: Research progress and future perspectives

Membrane technologies have become proficient alternatives for advanced wastewater treatment, ensuring high contaminant removal and sustainable resource recovery. Despite significant progress, ongoing research efforts aim to further optimize treatment performance. Among the challenges faced, membrane fouling persists as a relevant obstacle in membrane technologies, necessitating the development of more effective mitigation strategies. Mathematical models, widely employed for predicting treatment performance, generally exhibit low accuracy and suffer from uncertainties due to the complex and variable nature of wastewater. To overcome these limitations, numerous studies have proposed artificial intelligence (AI) modeling to accurately predict membrane technologies' performance and fouling mechanisms. This approach aims to provide advanced simulations and predictions, thereby enhancing process control, optimization, and intensification.This literature review explores recent advancements in modeling membrane-based wastewater treatment processes through AI models. The analysis highlights the enormous potential of this research field in enhancing the efficiency of membrane technologies. The role of AI modeling in defining optimal operating conditions, developing effective strategies for membrane fouling mitigation, enhancing the performance of novel membrane-based technologies, and improving membrane fabrication techniques is discussed. These enhanced process optimization and control strategies driven by AI modeling ensure improved effluent quality, optimized resource consumption, and minimized operating costs. The potential contribution of this cutting-edge approach to a paradigm shift toward sustainable wastewater treatment is examined. Finally, this review outlines future perspectives, emphasizing the research challenges that require attention to overcome the current limitations hindering the integration of AI modeling in wastewater treatment plants.

Feedback control of an experimental electrically-heated steam methane reformer

Steam methane reforming (SMR) is the most common industrial process to produce hydrogen (H2) from methane and water vapor. The SMR reactions are overall highly endothermic and traditionally, fossil fuels are burned to provide the necessary heat of reaction. However, it was found that electrifying the tubular SMR reactors using renewable electrons instead of conventional heating gives the opportunity of producing H2 with lower carbon emissions, lower reactor volumes, and higher carbon conversion. While current studies in this new technology focus on improving the catalysts or heating sources, a more efficient process control scheme can also significantly contribute to the understanding of potential challenges and opportunities of combining renewables and natural gas in the production of H2 at industrial scales. In order to further develop the electrically-heated SMR process and the associated control strategies, we have constructed an experimental Joule-heated SMR system at UCLA. The system contains a tubular reactor with a washcoated Ni/ZrO2 catalyst, two thermocouples connected to the top and bottom of the reactor, a power supply for providing electrical heating, and an on-line gas chromatograph (GC) for measuring outlet gas concentrations. The synthesis procedure of the catalyst, SMR data collection, and thermal considerations for catalyst degradation are described in this paper and are all accounted for experimentally when using a proportional integral (PI) controller to gradually increase the reactor temperature without harming the catalyst. Advanced control strategies, such as model predictive control (MPC), require a process model that uses measurement feedback to make predictions of the process time-evolution in order to optimize the control actions in real-time. We have previously developed a lumped-parameter modeling strategy for the SMR process. The MPC objective is to control the H2 production rate by manipulating the current flowing through the outer reactor shell. However, to use this model in an MPC, feedback values for all state variables should be provided by the sensors. In our experimental system, the on-line GC only gives discrete measurements with a long sampling period. To this end, the process model is incorporated into an extended Luenberger observer (ELO) that uses the reactor temperature and GC measurements to provide estimates of all the variables needed by the MPC. The ELO-based MPC system is then experimentally implemented on the process and it is demonstrated to be more efficient in terms of speed of the closed-loop response than the PI controller using delayed, measurement feedback by the GC.