Convolutional Neural Network (CNN)-Based Measurement of Properties in Liquid–Liquid Systems

Processes, Journal Year: 2023, Volume and Issue: 11(5), P. 1521 - 1521

Published: May 16, 2023

The rise of artificial intelligence (AI)-based image analysis has led to novel application possibilities in the field solvent analytics. Using convolutional neural networks (CNNs), better and more automated optically visible phenomena becomes feasible, broadening spectrum non-invasive measurements. These so-called smart sensors have attracted increasing attention pharmaceutical chemical process engineering; their additional sensor data enables precise control as parameters can be monitored. This contribution presents an approach analyzing single rising droplets determine physical properties; for example, geometrical such diameter, projection area volume. Additionally, velocity is determined, well density interfacial tension liquid droplet, determined from force balance. Thus, a method was developed liquid–liquid properties suitable real-time applications. Here, size range investigated droplet diameters lies between 0.68 mm 7 with accuracy AI detecting ±4 µm. obtained densities lie 0.822 kg·m−3 n-butanol 0.894 toluene droplets. For derived parameters, estimation, all points 12.75 mN·m−1 15.25 mN·m−1. trueness system thus −1 +0.4 mN·m−1, precision ±0.3 ±0.6 estimation using our system, standard deviation 1.4 kg m−3 literature determined. camera images conjunction improved by algorithms, combined empirical mathematical formulas, this article contributes development easily accessible, cheap sensors.

Language: Английский

---

Multiphase Flow’s Volume Fractions Intelligent Measurement by a Compound Method Employing Cesium-137, Photon Attenuation Sensor, and Capacitance-Based Sensor

Energies, Journal Year: 2024, Volume and Issue: 17(14), P. 3519 - 3519

Published: July 18, 2024

Multiphase fluids are common in many industries, such as oil and petrochemical, volume fraction measurement of their phases is a vital subject. Hence, there lots scientists researchers who have introduced methods equipment this regard, for example, photon attenuation sensors, capacitance-based so on. These approaches non-invasive reason, very popular widely used. In addition, nowadays, artificial neural networks (ANN) attractive lot fields because accuracy. Therefore, paper, to estimate proportion three-phase homogeneous fluid, new system proposed that contains an MLP ANN, standing multilayer perceptron network, sensor, sensor. Through computational methods, capacities mass coefficients obtained, which act inputs the network. All these were divided randomly two main groups train test presented model. To opt suitable network with lowest rate mean absolute error (MAE), number architectures different factors tested MATLAB software R2023b. After receiving MAEs equal 0.29, 1.60, 1.67 water, gas, phases, respectively, was chosen be paper. based on outcomes, approach’s novelty being able predict all flow low error.

Language: Английский

---

AI-Based Evaluation of Homogeneous Flow Volume Fractions Independent of Scale Using Capacitance and Photon Sensors

ARO-The Scientific Journal of Koya University, Journal Year: 2024, Volume and Issue: 12(2), P. 167 - 178

Published: Nov. 9, 2024

Metering fluids is critical in various industries, and researchers have extensively explored factors affecting measurement accuracy. As a result, numerous sensors methods are developed to precisely measure volume fractions multi-phase fluids. A significant challenge fluid pipelines the formation of scale within pipes. This issue particularly problematic petroleum industry, leading narrowed internal diameters, corrosion, increased energy consumption, reduced equipment lifespan, and, most crucially, compromised flow paper proposes non-destructive metering system incorporating an artificial neural network with capacitive photon attenuation address this challenge. The simulates thicknesses from 0 mm 10 using COMSOL multiphysics software calculates counted rays through Beer Lambert equations. simulation considers 10% interval variation each phase, generating 726 data points. proposed network, two inputs—measured capacity rays-and three outputs—volume gas, water, oil—achieves mean absolute errors 0.318, 1.531, 1.614, respectively. These results demonstrate system’s ability accurately gauge proportions three-phase gas-water-oil fluid, regardless pipeline thickness.

Language: Английский

---

Artificial Intelligence Techniques for the Hydrodynamic Characterization of Two-Phase Liquid–Gas Flows: An Overview and Bibliometric Analysis

Fluids, Journal Year: 2024, Volume and Issue: 9(7), P. 158 - 158

Published: July 8, 2024

Accurately and instantly estimating the hydrodynamic characteristics in two-phase liquid–gas flow is crucial for industries like oil, gas, other multiphase sectors to reduce costs emissions, boost efficiency, enhance operational safety. This type of involves constant slippage between gas liquid phases caused by a deformable interface, resulting changes volumetric fraction creation structures known as patterns. Empirical numerical methods used prediction often result significant inaccuracies during scale-up processes. Different methodologies based on artificial intelligence (AI) are currently being applied predict flow, which was corroborated with bibliometric analysis where AI techniques were found have been pattern recognition, determination each fluid, pressure gradient estimation. The results revealed that total 178 keywords 70 articles, 29 reached threshold (machine learning, pattern, intelligence, neural networks high predominance), published mainly Flow Measurement Instrumentation. journal has highest number articles related studied topic, nine articles. most relevant author Efteknari-Zadeh, E, from Institute Optics Quantum Electronics.

Language: Английский

---