Loading...
Search
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
Digital Object List
Bookmark
Review List
135 viewed
توسعه جریان سنج چندفازی مبتنی بر اطلاعات حسگر با استفاده از روش هوش مصنوعی
کاظمی، محمد حسین Kazemi, Mohammad Hosein
Development of the Multiphase Flowmeter Based on the Sensor Data using Artificial Intelligence Method
Kazemi, Mohammad Hosein | 2025
72
Viewed
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 58734 (08)
- University: Sharif University of Technology
- Department: Mechanical Engineering
- Advisor(s): Nejat Pishkenari, Hossein; Salarieh, Hassan
- Abstract:
- Multiphase flows involve the simultaneous passage of two or more phases (liquid, gas, or solid) and have widespread applications in the oil and gas, chemical, nuclear, biological, and food industries. Accurate measurement of each phase’s flow rate is crucial for improving equipment performance and ensuring industrial safety. This study focuses on the development of a water–air two-phase flowmeter based on pressure sensors and artificial intelligence models. Unlike traditional mathematical models, which require flow regime identification and complex physical relationships, deep neural networks are capable of learning nonlinear patterns from experimental data and handling environmental uncertainties. To enhance the accuracy and generalizability of the model, time–frequency features of the signals were extracted and the most effective ones were selected. The designed flowmeter system consists of a Venturi tube equipped with two differential pressure sensors and one gauge pressure sensor that record the inlet, throat, and outlet pressures. Data acquisition was conducted under diverse operating conditions, and the signals were filtered using a digital filter. Time–frequency features were extracted, and data volume was increased using 30-second sliding windows. Subsequently, a multilayer perceptron neural network was developed, and the effects of different feature selection methods (correlation, variance threshold, principal component analysis, Lasso regularization, and gradient boosting) were investigated. The results indicated that the best performance was achieved with Lasso regularization and 10 selected features, yielding a mean relative error of 11.47% for water flow rate and 19.94% for air flow rate, whereas principal component analysis with 10 components resulted in errors of 16.1% for water and 27.3% for air. This research represents an important step toward the development of indigenous multiphase flowmeters based on simple hardware and artificial intelligence methods
- Keywords:
- Multiphase Flowmeter ; Pressure Measurement Sensor ; Sensor Data Fusion ; Artificial Intelligence Methods ; Water and Air Two Phase Flow ; Time–Frequency Feature Extraction ; Venturi Tube Flowmeter
-
محتواي کتاب
- view