Machine learning applications to predict two-phase flow patterns

Abstract

Recent advances in artificial intelligence with traditional machine learning algorithms and deep learning architectures solve complex classification problems. This work presents the performance of different artificial intelligence models to classify two-phase flow patterns, showing the best alternatives for this specific classification problem using two-phase flow regimes (liquid and gas) in pipes. Flow patterns are affected by physical variables such as superficial velocity, viscosity, density, and superficial tension. They also depend on the construction characteristics of the pipe, such as the angle of inclination and the diameter. We selected 12 databases (9,029 samples) to train and test machine learning models, considering these variables that influence the flow patterns. The primary dataset is Shoham (1982), containing 5,675 samples with six different flow patterns. An extensive set of metrics validated the results obtained. The most relevant characteristics for training the models using Shoham (1982) dataset are gas and liquid superficial velocities, angle of inclination, and diameter. Regarding the algorithms, the Extra Trees model classifies the flow patterns with the highest degree of fidelity, achieving an accuracy of 98.8%.

Multiscale Turbulence Characteristics of Gas–Liquid Pressure Fluctuations in an Industrial Quatro Static Mixer

“…Machine learning models are established based on training data and are not influenced by physical constraints. The flow pattern of the gas–liquid two-phase was mainly identified by visualization research in the past, but it could not accurately capture the subtle variation in the flow pattern. , In recent years, machine learning methods have been introduced into the flow pattern identification of the gas–liquid two-phase, which could provide a reference for the advancement of flow pattern recognition technology. , The approximation solutions and higher flow pattern recognition accuracy can be obtained by using ML mechanisms for flow pattern recognition without simulation or a training data set. , Figure depicts the confusion matrix of the flow pattern recognition results using SVM and RF. There are 128 working conditions in the experiment.…”

Requests classification in the customer service area for software companies using machine learning and natural language processing

“…The chosen ML techniques that stand out for solving classification tasks are: Logistic regression (LR), linear discriminant Analysis (LDA), decision trees (DT), Random Forests (RF), Extra Trees (ET), k-nearest neighbors (KNN), support vector machines (SVM), and naïve Bayes(NB) ( Kotsiantis, Zaharakis & Pintelas, 2006 ; Osisanwo et al., 2017 ; Arteaga-Arteaga et al., 2021 ).…”

“…Evaluating machine learning algorithms is an essential part of any project. The model may perform well when evaluated against one metric but poorly assessed against other metrics ( Hossin & Sulaiman, 2015 ; Arteaga-Arteaga et al., 2021 ).…”

“…Accuracy : is the total number of input samples ratio to the number of correct predictions ( Hossin & Sulaiman, 2015 ; Arteaga-Arteaga et al., 2021 ): …”

“…Precision : this metric allows for determining the number of correct positive predictions made ( Hossin & Sulaiman, 2015 ; Arteaga-Arteaga et al., 2021 ). …”