Temporal Data and Short-Time Averages Improve Multiphase Mass Flow Metering
🤖 AI Summary
Accurate mass flow measurement under multiphase flow conditions—such as gas–water–oil mixtures—remains challenging for conventional instrumentation. This work proposes a novel approach that integrates machine learning with Coriolis mass flowmeter data, systematically demonstrating for the first time that a short-time averaging strategy preserving temporal structure outperforms traditional single-point averaging, thereby significantly enhancing model generalization and robustness. Using multilayer perceptrons (MLPs), sliding-window MLPs, and convolutional neural networks (CNNs) to process time-series data at varying downsampled frequencies, experiments show that the CNN achieves optimal performance at 0.25 Hz, yielding a 95th-percentile relative error below 13%, a normalized root mean square error of 0.03, and a mean absolute percentage error of approximately 4.3%.
Application Category
📝 Abstract
Reliable flow measurements are essential in many industries, but current instruments often fail to accurately estimate multiphase flows, which are frequently encountered in real-world operations. Combining machine learning (ML) algorithms with accurate single-phase flowmeters has therefore received extensive research attention in recent years. The Coriolis mass flowmeter is a widely used single-phase meter that provides direct mass flow measurements, which ML models can be trained to correct, thereby reducing measurement errors in multiphase conditions. This paper demonstrates that preserving temporal information significantly improves model performance in such scenarios. We compare a multilayer perceptron, a windowed multilayer perceptron, and a convolutional neural network (CNN) on three-phase air-water-oil flow data from 342 experiments. Whereas prior work typically compresses each experiment into a single averaged sample, we instead compute short-time averages from within each experiment and train models that preserve temporal information at several downsampling intervals. The CNN performed best at 0.25 Hz with approximately 95 % of relative errors below 13 %, a normalized root mean squared error of 0.03, and a mean absolute percentage error of approximately 4.3 %, clearly outperforming the best single-averaged model and demonstrating that short-time averaging within individual experiments is preferable. Results are consistent across multiple data splits and random seeds, demonstrating robustness.
Problem
Research questions and friction points this paper is trying to address.
multiphase flow
mass flow metering
measurement error
temporal data
short-time averages
Innovation
Methods, ideas, or system contributions that make the work stand out.
temporal data
short-time averaging
multiphase flow metering
convolutional neural network
Coriolis mass flowmeter
🔎 Similar Papers
No similar papers found.
