Industrial Process Control Research Articles

Assessment and deployment of a LSTM-based virtual sensor in an industrial process control loop

AbstractMeasurement of certain variables within the industrial sector remains a challenge due to the prohibitive costs of sensors, the intricate installation processes, or the continuous nature of production demands. Moreover, if a backup sensor is required in case the main sensor fails, the installation and maintenance difficulties are further increased. A possibility to address this issue is the indirect estimation of the desired variable by leveraging other correlated measures within the operational process. Data-driven techniques are well-suited for this aim, given their capacity to model potentially complex industrial processes. This paper proposes the implementation of a virtual flow sensor for its integration in the control loop of an industrial process. More specifically, four different data-driven methods have been tested to obtain the virtual sensor: multiple linear regression (MLR), multilayer perceptron (MLP), long-short term memory (LSTM) and deep long-short term memory (DeepLSTM). MAE, RMSE and $$R^2$$ R 2 have been chosen as evaluation metrics for model selection and testing. Furthermore, the robustness of the virtual flow sensor is not only evaluated under ideal operating conditions, but it is also tested under adverse conditions with various noise levels added to the measured signals. Additionally, the performance of the flow control loop using the real and virtual sensors is also evaluated in both ideal and adverse conditions. IAE, ITAE, and IAVU indices are used to assess the control performance. The results prove the robustness of the LSTM-based virtual flow sensor and the effectiveness of the control loop using it, avoiding the modification of the controller and interrupting the process when the real flow sensor fails.

LIBS as a fast and reliable alternative to µXRF and SEM–EDX for quantitative analysis of aluminium alloy particles in technical cleanliness analysis

The characterization of metal alloy (particle) characterization is a vital tool in today’s high–tech applications for quality and process control in industry, especially for applications in technical cleanliness analysis (TCA). Here, metal alloy fragments down to sizes of 200 µm need to be quantitatively characterized as accurate as possible. Laser-induced breakdown spectroscopy (LIBS) offers several advantages due to its easy use, short measurements times, and high sensitivity. To enable the application of LIBS in this context, the method must yield adequately accurate results. Therefore, in this study we assess the performance of LIBS by comparison with well–established methods – scanning electron microscopy coupled with energy-dispersive X‑ray spectroscopy (SEM-EDX) and µ‑X‑ray fluorescence (µXRF). We analyse particulate certified reference materials consecutively with non–destructive and destructive methods on the quantitative composition regarding the key elements necessary for distinguishing aluminium alloys: Si, Zn, Mn, Mg, Cu and Fe. Sample inhomogeneity and the effect of sampling location and area is shown via microstructure analysis and combined with an assessment of the analysis volume for each used method. By evaluating the data statistically and by comparing quantitative results achieved with different methods, we can postulate a high agreement between LIBS and the studied reference methods, providing a basis for the use of LIBS in industrial analysis of metal alloy particles.

A temporal convolution network-based just-in-time learning method for industrial quality variable prediction

Real-time acquisition of quality variables is paramount for enhancing control and optimization of industrial processes. Process modeling methods, such as soft sensors, offer a means to predict difficult-to-obtain quality variables using easily measurable process parameters. However, the dynamic nature of industrial processes poses significant challenges to modeling. For instance, conventional models are typically trained offline using historical data, rendering them incapable of adapting to real-time changes in data distribution or environmental conditions. To tackle this challenge, we introduce a novel approach termed the Residual Temporal Attention Temporal Convolution Network (RTA-TCN) and propose a just-in-time learning method based on RTA-TCN for industrial process modeling. The RTA-TCN model incorporates temporal attention into TCN, enabling the integration of previous time-step process variables into the current ones, as well as the fusion of internally relevant features among inputs. Moreover, to prevent the partial loss of original information during feature integration, residual connections are introduced into the temporal attention mechanism. These connections facilitate the retention of original feature information to a maximal extent while integrating relevant features. Consequently, the proposed RTA-TCN demonstrates significant advantages in handling the non-linearity and long-term dynamic dependencies inherent in industrial variables. Additionally, the proposed just-in-time learning method leverages RTA-TCN as a local model and updates it in real-time using online industrial data. This just-in-time learning method enables effective adaptation to varying data distributions and environmental conditions. We validate the performance of our method using two industrial datasets (Debutanizer Column and Sulfur Recovery Unit).

Hybrid Model using Bond Graph-TCN Network and Event Triggered Predictive Control of pH Neutralization Process

The pH neutralization process is a concrete decisive unit in major reactor units of industrial process control loops. This article presents a new fuzzy-based hybrid 'Bond Graph-Temporal Convolution Network' (BG-TCN) model, structured for the convoluted dynamics of a real-time pH neutralization unit, known for its complexity and high nonlinearity. The TCN scheme suggested in this article, exploits a one-dimensional causal convolution strategy within a residual learning framework to execute dilated causal convolutions through time series data analysis. Conversely, Bond Graph (BG) is a graphical tool, designed on an energy-centric approach, to represent energy transfer and interactions across different compartments of the nonlinear pH neutralization system. Furthermore, a linguistic fuzzy rule-based inference system is encompassed to handle uncertainties from BG and TCN models, allowing smooth integration and flexible transition between these two approaches. Additionally, the performance of the hybrid BG-TCN model is assessed against the individual TCN and BG models in a Python environment. On top of that, this article also envisions an event-triggered predictive control utilizing a fuzzy event handler mechanism to demonstrate the efficacy of the proposed hybrid BG-TCN in attaining precise set point tracking for closed-loop servo and regulatory problems.

Fault diagnosis of pneumatic control valve based on Bat Optimization algorithm and BP neural network

Abstract This article proposes a fault diagnosis model for pneumatic control valves based on a bat optimization (BA) algorithm and backpropagation (BP) neural network. The DAMADICS platform simulates the pneumatic industrial process control valve failures, and the fault signals are collected to construct the data set. A fault diagnosis model combining the Bat Optimization algorithm and Back Propagation neural network (BA-BP) is built, and four common faults are injected for experiments. The method is compared with Support Vector Machine (SVM), BP, and Bayesian optimized BP (BO-BP). Experimental results indicate that BA-BP has high diagnostic precision in the diagnosis of malfunctions in pneumatic valves.

Design and Investigation of a High-Performance Quartz-Based SAW Temperature Sensor.

In this work, a surface acoustic wave (SAW) temperature sensor based on a quartz substrate was designed and investigated. Employing the Coupling-of-Modes (COM) model, a detailed analysis was conducted on the effects of the number of interdigital transducers (IDTs), the number of reflectors, and their spacing on the performance of the SAW device. The impact of the transversal mode of quartz SAWs on the device was subsequently examined using the finite element method (FEM). The simulation results indicate that optimizing these structural parameters significantly enhances the sensor's sensitivity and frequency stability. SAW devices with optimal structural parameters were fabricated, and their resonant frequencies were tested across a temperature range of 25-150 °C. Experimental results demonstrate that the SAW temperature sensor maintains high performance stability and data reliability throughout the entire temperature range, achieving a Bode-Q of 7700. Furthermore, the sensor exhibits excellent linearity and repeatability. An analysis of the sensor's response under varying temperature conditions reveals a significant temperature dependency on its Temperature Coefficient of Frequency (TCF). This feature suggests that the sensor possesses potential advantages for applications in industrial process control and environmental monitoring.

Enhancing DC Motor Speed Control Performance Using Heuristic Optimization and Comparative Analysis of Control Methods

Direct Current (DC) motors are an important component that converts electrical energy into mechanical energy, used in a wide range of applications from industrial applications to home appliances. DC motor speed control has an important role in industrial processes to increase efficiency, realize precise movements and optimize energy consumption. In this study, various control methods and parameter optimization techniques for speed control of DC motors, which have a wide range of applications, have been systematically analyzed. The aim of the study is to develop an effective control strategy to ensure that DC motors reach the determined target speed by monitoring them in real time at different speeds and to minimize fluctuations caused by variable loads or external factors. In our study, Proportional-Integral-Derivative (PID), Proportional-Integral (PI), and Proportional-Derivative (PD) control methods were used. The parameters of these controllers were tuned using Matlab Tuned, The Cheetah Optimizer (CO) Algorithm, a new generation heuristic optimization method, and Particle Swarm Optimization (PSO), a widely accepted optimization method. The performances of the controllers were determined using criteria such as Integral of Absolute Error (IAE), Integral Squared Error (ISE), and Integral of Time multiplied by Absolute Error (ITAE). According to the results obtained, it was found that the PID, PI and PD control parameters determined using the CO Algorithm performed better than the controllers created using Matlab Tuned and PSO methods. New optimization methods, such as the CO Algorithm, have been found to have significant potential to improve the performance of control systems. Thanks to this study, it offers a practical approach for optimizing DC motor speed control in industrial processes. As a result, it has been found that the control parameters determined by the CO Algorithm have significant potential in improving the performance of DC motor speed control and control systems compared to other optimization methods.

Review on Application of Sustainability, Circular and Digital Economy on Bioplastics Production

AbstractDigitalization is creating and driving a sustainable data driven production and consumption of materials and energy. The goal of this research was to outline circular economy and employ machine learning algorithms in the industry 4.0-environmental social and governance (ESG) for the bioplastic properties. The paper review sustainability, circular and digital economy and potential of biomasses on bioplastics production, tensile strength and degradation. Sustainability and circular economy are very crucial to adaptation and mitigation of climate change and social-economic responsibilities of the environmental and human health. It was found that most studies covered sustainability, few embed circular and digital economy. It was important to investigate the end-of life of a product, to know its effect to the environment and human health. There were limited studies on bioplastics production with the digital economy. It was discovered that machine learning approaches have the potential to improve quality control and optimization in industrial processes.

Conductive metal-organic framework for ppb-concentration and highly selective SAW hydrogen sulfide sensing at room temperature

Hydrogen sulfide (H2S) sensing is utilized for monitoring the concentration in environment protection and industrial process control, of which high selectivity and ppb detection limit are required. Here, a ppb-concentration and highly selective surface acoustic wave (SAW) H2S sensing has been developed with the Cu3(HHTP)2 MOFs conductive metal-organic frameworks (MOFs), which have been synthesized by connecting Cu2+ with the organic ligands via hydrothermal process. Typically, as-prepared Cu3(HHTP)2 MOFs take on shapes including nanoparticles and nanorods. Beneficially, the SAW H2S sensing presents excellent selectivity to H2S and enables to detection of H2S as low as 6 ppb at room temperature (∼ 26 °C). Remarkably, the SAW sensor prototypes exhibit high sensitivity (0.02 mV/ppb) to 50–2000 ppb H2S, fast response (T90: 281 s) and 46 day-long stability at room temperature. Theoretically, such excellent H2S SAW sensing performance might be attributed to the interaction between Cu ions and H2S molecules and the strong acoustoelectric effect of SAW. Practically, our ppb-concentration and highly selective H2S SAW sensing have the potential in the real-time detection of H2S.

Integrative approach to fluorescence-based oxygen sensing in polymer optical fibers with surface plasmon resonance

This study demonstrates a technique to enhance oxygen (O2) concentration measurement by combining surface plasmon resonance (SPR) and fluorescence-based sensor on polymer optical fiber (POF). The SPR was generated using silver (Ag) and the fluorescence was generated by organic dye. Three fluorescence dyes were used, Tris (4, 7-diphenyl-1, 10-phenanthroline) ruthenium (II) dichloride ([Ru(dpp)3]2+), platinum octaethylporphyrin (PtOEP) and 5,10,15,20-tetrakis (pentafluorophenyl) 21 H,23H-porphine palladium (II) (PdTFPP). The sensor was tested in a gas chamber with different concentration level of O2. Sensitivity values for the fluorescence dyes for [Ru(dpp)₃]²⁺, PtOEP, and PdTFPP were 0.0099 a.u/%, 0.0343 a.u/% and 0.05 a.u/% respectively. When SPR was implemented, the sensitivity values for Ag/[Ru(dpp)₃]²⁺, Ag/PtOEP, and Ag/PdTFPP were 0.0589 nm/%, 0.0345 nm/%, and 0.118 nm/% respectively. PdTFPP and Ag/PdTFPP exhibit highest sensitivity in each experiment. Therefore, the further analysis for the sensing performance of PdTFPP and Ag/PdTFPP were held. As a result, Ag/PdTFPP exhibits lower limit of detection (LOD), and limit of quantification (LOQ) with 3.054 % and 9.254 %, respectively, with higher R2 of 0.9971 and lower mean standard deviation of 0.174, compared to PdTFPP 16.496 % and 49.988 % respectively, with R2 of 0.9211 and higher mean standard deviation of 2.003. This improvement can benefit researchers and professionals in medical diagnostics, environmental monitoring, and industrial process control by providing a more sensitive and accurate method for measuring O2 levels.

Application of digital twin for industrial process control: A case study of gauge-looper-tension optimized control in strip hot rolling

Background During the hot rolling process, the performance of most control systems gradually degrades due to equipment aging and changing process conditions. However, existing gauge-looper-tension control method remain restricted by a lack of intelligent parameter maintenance strategies. Methods To address this challenge and enhance the smart manufacturing capabilities of strip hot rolling, based on the digital twin method, this paper proposes a data-driven optimized control method for the gauge-looper-tension system of strip hot rolling. First, a hot rolling process model is constructed based on a digital twin method to serve as an evaluation and optimization platform. Subsequently, relevant data are collected to calculate the Hurst index for identifying the performance of the controller during the rolling process. Additionally, for controllers with poor Hurst index values, the crayfish optimization algorithm is employed for adjusting controller parameters to maximum the Hurst index. Results A real case of hot rolling steel production was used to validate the proposed data-driven optimized control method. Experimental results demonstrate that the proposed evaluation method can effectively recognize the state of gauge-looper-tension controller. Moreover, after optimizing the controller parameters through crayfish optimization algorithm (COA), the value of Hurst index showed significant improvement. Conclusions The proposed digital twin -based optimized control method effectively enhance the manufacturing capability and maintain strategy of hot rolling steel production. Through the proposed method, the Hurst index of gauge-looper-tension system increasing from 0.574 to 0.862.

Optimizing olefin purification: An artificial intelligence-based process-conscious PI controller tuning for double dividing wall column distillation

This study investigates the light olefin separation using dual dividing wall columns (DWCs) and compares various proportional-integral (PI) controller tuning methods. A novel process-conscious reinforcement learning (RL) approach, utilizing Deep Deterministic Policy Gradient (DDPG), is developed by incorporating domain-specific knowledge into the reward function to ensure compliance with critical process variables and product purity. The DDPG-based PI tuning is evaluated against traditional methods such as Aspen’s recommended initial tuning, Ziegler-Nichols (ZN), and Cohen-Coon (CC). The result shows that the DDPG method significantly enhances control stability and accuracy, reducing error by factors of 11.9, 2.3, and 1.6 compared to Aspen, ZN, and CC, respectively. Additionally, DDPG demonstrates superior energy efficiency, consuming 13% less energy than the next best method. It also reduces purification costs by up to 13.5% and CO2 emissions by 20% compared to other methods. This study highlights the potential of integrating advanced RL techniques into industrial process control, delivering substantial improvements in stability, energy efficiency, economic, and environmental performance.

Manganese Oxide Applications in Sulfonamides Electrochemical, Thermal and Optical Sensors: A Short Review

AbstractIn recent years, the development of highly sensitive and selective electrochemical sensors has been a pivotal area of research, driven by the growing demand for environmental monitoring and industrial process control. Among various materials investigated for sensor applications, manganese oxide (MnO2) nanoparticles have garnered significant attention due to their excellent electrochemical properties, environmental friendliness, and natural abundance. Critical analyses of the synthesis of MnO2 using different techniques such as hydrothermal method, chemical precipitation, and sol–gel process which allows for the fine-tuning of particle size and morphology while enhancing the electrochemical sensing capabilities have been reviewed. The review also provides a comprehensive overview of the recent advancement evaluation of manganese oxide-based electrodes for detecting sulfonamides and other analytes in water across diverse matrices. This paper sets the stage for a comprehensive exploration of the synthesis methods and application areas of MnO2 nanoparticles in electrochemical sensors, highlighting their role in advancing sensor technology and their impact on various sectors. Graphical Abstract

Soft sensor modeling based on self-organizing fuzzy neural network with clustering, merging, and splitting scheme

Abstract As a pivotal role in the control, optimization, and monitoring of contemporary industrial processes, soft sensors are frequently employed in the prediction of key quality variables. To achieve accurate prediction of key quality variables in industrial processes, a soft sensor modeling method based on the self-organizing fuzzy neural network with the clustering, merging, and splitting scheme (SOFNN-CMS) is proposed. First, the supervised fuzzy C-means clustering algorithm is proposed to identify the appropriate initial center and width of the fuzzy neural network, obtaining appropriate initial fuzzy rules. Then, a neuron merging and splitting strategy is designed to adjust the structure of the fuzzy neural network, by merging and splitting the hidden neurons according to the distance of clusters, increasing the adaptability of the fuzzy neural network. Besides, to accelerate the convergence of estimation errors, an improved Levenberg Marquardt algorithm is utilized to update neural network parameters in the training phase, realizing the soft sensor modeling of key quality variables. The effectiveness of the proposed SOFNN-CMS neural network is demonstrated on two benchmark problems and an industrial debutanizer column. Finally, the experiments showcase that the proposed SOFNN-CMS neural network can obtain better soft sensor modeling performance with a compact structure.

Review of potentiometric determination of cationic surfactants

Abstract Cationic surfactants (CSs) are surface-active compounds containing a positively charged polar group and at least one alkyl chain as a nonpolar group. Due to their structure, they tend to adsorb on negatively charged surfaces and interact with biopolyanions. It leads to their wide use as disinfectants, cleaning agents, fabric softeners, hair care products, emulsifiers, corrosion inhibitors, etc. Considering their extensive use and also their toxicity, fast, simple, and accurate CSs determination is crucial in industrial process control, product quality assurance, and environmental monitoring. Potentiometric sensors meet all these requirements, so they stand as the primary method for CSs determination. In this review, numerous potentiometric methods for CSs determination have been described, with a particular focus on methods published in the period from 2000 to 2024. Due to their simplicity and good analytical performance, solid-state electrodes are the most commonly used type of sensor for CSs determination.

The Artificial Neural Twin — Process optimization and continual learning in distributed process chains

Industrial process optimization and control is crucial to increase economic and ecologic efficiency. However, data sovereignty, differing goals, or the required expert knowledge for implementation impede holistic implementation. Further, the increasing use of data-driven AI-methods in process models and industrial sensory often requires regular fine-tuning to accommodate distribution drifts. We propose the Artificial Neural Twin, which combines concepts from model predictive control, deep learning, and sensor networks to address these issues. Our approach introduces decentral, differentiable data fusion to estimate the state of distributed process steps and their dependence on input data. By treating the interconnected process steps as a quasi neural-network, we can backpropagate loss gradients for process optimization or model fine-tuning to process parameters or AI models respectively. The concept is demonstrated on a virtual machine park simulated in Unity, consisting of bulk material processes in plastic recycling.

Feedback system for reactor process analysis

The article discusses an approach to analyzing processes in industrial reactors using advanced systems with feedback. The authors describe the system preparation and image reconstruction algorithms of ultrasonic tomographs (USTs) that have applications in industry. The research presented in this paper focuses on developing effective strategies for controlling and monitoring chemical reaction processes in reactors. Using advanced data processing techniques, the authors propose systems with feedback that enable accurate analysis and optimization of reactor operating conditions. A key aspect is ultrasonic tomographic imaging, which provides precise data on the state of the process in the reactor. In summary, the paper presents a state-of-the-art approach to analyzing chemical reaction processes in industrial reactors using advanced feedback systems and ultrasonic imaging technologies. The proposed solutions are essential for improving efficiency and process control in the chemical industry.

Generalized collaborative relevance vector regression for soft sensors

Soft sensors often struggle to adapt to changing conditions due to their overreliance on single data sources, severely undermining the generalizability and applicability of models. This paper proposes a novel generalized collaborative relevance vector regression (GCRVR) framework to substantially improve the accuracy and adaptability of soft sensors under varying conditions. The core innovation of GCRVR lies in an adaptive collaborative strategy designed to combine complementary predictive distributions based on analyzed variance differences, enabling the acquisition of more reasonable and reliable results. The GCRVR approach leverages both smoothed and unsmoothed process data to train independent relevance vector regression models, generating probabilistic predictions. A theoretical analysis provides insight into ensuring the stability of the model. In an evaluation conducted across four sophisticated industrial processes, the proposed GCRVR method outperformed a suite of conventional methods—namely, linear regression, support vector regression, Gaussian process regression, a stacked autoencoder, long short-term memory, attention-based long short-term memory, and extreme gradient boosting. The GCRVR method exhibited substantial predictive accuracy enhancements, as evidenced by average reductions of 39% in the root mean square error (RMSE), 43% in the mean absolute error (MAE), and 42% in the mean absolute percentage error (MAPE). This work addresses the generalization and adaptability limitations of soft sensors through an integrated framework that capitalizes on the collaborative use of complementary models. GCRVR constitutes a major advancement in real-time monitoring and optimization for industrial process control.

High-Performance Humidity Sensor Capable for Monitoring Low-Moisture Levels in Energy Industries

Humidity sensors are extensively used in industrial processing and environmental control. There are two categories of humidity sensors: High-humidity (relative humidity) measurement and Low-humidity (moisture) measurement. The proper units for low-humidity measurement are dew point and parts per million in volume (ppmv) or parts per billion in volume (ppbv). It is very important to monitor moisture levels in natural gas during processing and transportation in pipelines. It is of critical importance to monitor moisture levels in lithium-ion battery manufacturing. In these cases, the moisture levels are limited to be less than -50°C dew point or 40 ppmv. The conventional relative humidity sensors cannot be used because of their low sensitivity. The dew point sensors on the market are polymer-based and gamma-phase-alumina based. The polymer-based sensors have low sensitivity, which can only accurately detect moisture levels of -60°C dew point and higher. The gamma-phase alumina sensors suffer from long-term drift due to phase change.We present novel high-performance humidity sensors based on porous alpha-phase alumina and silicon dioxide nanostructures, which are capable of detecting moisture levels as low as -100°C dew point or 13 ppbv with excellent long-term stability. The porous alpha-phase alumina films were produced by anodic-spark-deposition on aluminum substrates in which electric sparks were produced due to high-voltage anodization (>120V). Local high-temperature (about 2200°C) at the points of sparks converted the amorphous alumina into alpha-phase alumina (Sapphire). It provides an extremely stable structure for humidity sensors. Hydrophilic silicon dioxide films deposited by atomic layer deposition was used as sensing materials. The sensors’ sensitivity, temperature coefficient, response speed, and long-term stability were studied in details. The sensors have great promise for monitoring moisture levels in natural gas industry and lithium-ion battery manufacturing.

Colorimetric IPN hydrogels embedded with colloidal photonic crystals: A novel approach for the detection of ethanol and Ba2+ ions in water

A critical bottleneck in sensor technology is the rapid and precise detection of specific analytes in complex matrices, hindering advancements in environmental monitoring, healthcare, and industrial process control. This study addresses this challenge by introducing a novel composite hydrogel sensor designed for rapid and selective detection of ethanol and barium ions (Ba2+) in aqueous environments. The sensor integrates interpenetrating network (IPN) hydrogels with embedded colloidal photonic crystals (CPCs), synthesized via a solution-based polymerization approach. This innovative configuration allows CPCs to dynamically adjust their photonic bandgap in response to environmental changes, manifesting as a visible, colorimetric shift. This response stems from the synergy between the mechanical properties of the IPN hydrogel and the optical sensitivity of CPCs. Upon exposure to analytes such as ethanol and Ba2+, the sensor exhibits a rapid and reversible color transition that is directly proportional to their concentration. Notably, ethanol (0 vol%–80 vol%) and Ba2+ (5–17.5 mM) induce a distinct blueshift in the photonic bandgap and trigger a color change from red–orange to green due to the alteration in the swelling behavior of the IPN hydrogel, affecting its lattice constant. The IPN hydrogel–CPC composite demonstrates exceptional operational stability and facilitates rapid detection, making it ideal for on-site applications without the need for complex equipment. These characteristics make the composite hydrogel sensor a promising candidate for environmental monitoring, industrial process control, and public health diagnostics, paving the way for the development of next-generation responsive sensor materials.