A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis
Flow regime classification is essential for analyzing and modeling two-phase flows, as it demarcates the flow behavior and influences the selection of appropriate predictive models. Machine learning-based approaches have gained relevance in flow regime classification research in the last few years....
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10840192/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832590334960336896 |
|---|---|
| author | Jefferson Dos Santos Ambrosio Marco Jose da Silva Andre Eugenio Lazzaretti |
| author_facet | Jefferson Dos Santos Ambrosio Marco Jose da Silva Andre Eugenio Lazzaretti |
| author_sort | Jefferson Dos Santos Ambrosio |
| collection | DOAJ |
| description | Flow regime classification is essential for analyzing and modeling two-phase flows, as it demarcates the flow behavior and influences the selection of appropriate predictive models. Machine learning-based approaches have gained relevance in flow regime classification research in the last few years. However, they are still solidly based on the construction and careful definition of hand-crafted features. Deep learning approaches, on the other hand, can provide more robust and end-to-end solutions. However, they are underexplored and have not evaluated the generalization of the models to other data or acquisition systems. Hence, this work proposes using end-to-end state-of-the-art (SOTA) time-series classification methods (ResNet, LSTM-FCN, and TSTPlus) for two-phase flow patterns (churn, bubbly, and slug). We also present the generalization analysis of the models with cross-dataset experiments, training the model with one dataset and testing it with another dataset collected in another system for two datasets: HZDR (from the Helmholtz-Zentrum Dresden-Rossendorf research laboratory) and TUD (from Technische Universität Dresden). The results demonstrate that the approach chosen here presents superior classification metrics in all cases evaluated, particularly in cross-dataset experiments. With our proposed SOTA methods, all the evaluated metrics (accuracy and F1-Score) consistently surpass 85% in all cases, while the baseline method can decrease the performance under 75%. This demonstrates the relevance of the analysis proposed here for flow regime classification literature and opens up a new set of possibilities for research in this area, aiming at robust solutions that are viable for practical use. Codes are available at <uri>https://github.com/ambrosioj/two-phase-time-series-deep-learning</uri>. |
| format | Article |
| id | doaj-art-714688df840c4fbb96a65807870ab734 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-714688df840c4fbb96a65807870ab7342025-01-24T00:01:37ZengIEEEIEEE Access2169-35362025-01-0113117781179110.1109/ACCESS.2025.352947210840192A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series AnalysisJefferson Dos Santos Ambrosio0https://orcid.org/0000-0002-6271-2236Marco Jose da Silva1https://orcid.org/0000-0003-1955-8293Andre Eugenio Lazzaretti2https://orcid.org/0000-0003-1861-3369Graduate Program in Electrical Engineering and Industrial Informatics (CPGEI), Universidade Tecnológica Federal do Paraná (UTFPR), Curitiba, BrazilInstitute of Measurement Technology, Johannes Kepler University Linz, Linz, AustriaGraduate Program in Electrical Engineering and Industrial Informatics (CPGEI), Universidade Tecnológica Federal do Paraná (UTFPR), Curitiba, BrazilFlow regime classification is essential for analyzing and modeling two-phase flows, as it demarcates the flow behavior and influences the selection of appropriate predictive models. Machine learning-based approaches have gained relevance in flow regime classification research in the last few years. However, they are still solidly based on the construction and careful definition of hand-crafted features. Deep learning approaches, on the other hand, can provide more robust and end-to-end solutions. However, they are underexplored and have not evaluated the generalization of the models to other data or acquisition systems. Hence, this work proposes using end-to-end state-of-the-art (SOTA) time-series classification methods (ResNet, LSTM-FCN, and TSTPlus) for two-phase flow patterns (churn, bubbly, and slug). We also present the generalization analysis of the models with cross-dataset experiments, training the model with one dataset and testing it with another dataset collected in another system for two datasets: HZDR (from the Helmholtz-Zentrum Dresden-Rossendorf research laboratory) and TUD (from Technische Universität Dresden). The results demonstrate that the approach chosen here presents superior classification metrics in all cases evaluated, particularly in cross-dataset experiments. With our proposed SOTA methods, all the evaluated metrics (accuracy and F1-Score) consistently surpass 85% in all cases, while the baseline method can decrease the performance under 75%. This demonstrates the relevance of the analysis proposed here for flow regime classification literature and opens up a new set of possibilities for research in this area, aiming at robust solutions that are viable for practical use. Codes are available at <uri>https://github.com/ambrosioj/two-phase-time-series-deep-learning</uri>.https://ieeexplore.ieee.org/document/10840192/Wire-mesh sensordeep learninggas-liquid two-phase flowtransformerflow pattern |
| spellingShingle | Jefferson Dos Santos Ambrosio Marco Jose da Silva Andre Eugenio Lazzaretti A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis IEEE Access Wire-mesh sensor deep learning gas-liquid two-phase flow transformer flow pattern |
| title | A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis |
| title_full | A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis |
| title_fullStr | A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis |
| title_full_unstemmed | A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis |
| title_short | A Deep Learning-Based Approach for Two-Phase Flow Pattern Classification Using Void Fraction Time Series Analysis |
| title_sort | deep learning based approach for two phase flow pattern classification using void fraction time series analysis |
| topic | Wire-mesh sensor deep learning gas-liquid two-phase flow transformer flow pattern |
| url | https://ieeexplore.ieee.org/document/10840192/ |
| work_keys_str_mv | AT jeffersondossantosambrosio adeeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis AT marcojosedasilva adeeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis AT andreeugeniolazzaretti adeeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis AT jeffersondossantosambrosio deeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis AT marcojosedasilva deeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis AT andreeugeniolazzaretti deeplearningbasedapproachfortwophaseflowpatternclassificationusingvoidfractiontimeseriesanalysis |