Machine Learning

Anomaly Detection in Multiphase Flow

Python ClassificationFeature EngineeringAnomaly Detection

Anomaly Detection in Multiphase Flow - project visual

The Problem

Multiphase air-water flow facilities need early detection of faults (air leakage and air blockage) to avoid operational and safety issues.

Approach

Used 29 sensor variables to build an ML-based anomaly detection system: data cleaning, feature engineering, and model comparison to distinguish normal vs abnormal operation and classify the two fault types.

Results

Improved detection accuracy for air leakage and air blockage, enabling faster response to abnormal conditions in the facility.

Key result: Classified 2 fault types across 29 sensors

Tools Used

PythonScikit-learnPandasMatplotlib

GitHub - Coming Soon