A Method for Statistical Processing of Magnetic Field Sensor Signals for Non-Invasive Condition Monitoring of...

Area of Science:

• Electrical Engineering
• Mechanical Engineering
• Signal Processing

Background:

• Condition monitoring of synchronous generators is crucial for reliable power system operation.
• Non-invasive techniques are preferred to avoid operational interference.
• External magnetic field monitoring offers a promising avenue for early fault detection.

Purpose of the Study:

• To propose a low-cost, computationally efficient strategy for synchronous generator condition monitoring and fault detection.
• To extract diagnostic information from the magnetic signature of synchronous generators.
• To develop an automated system for continuous monitoring and anomaly detection.

Main Methods:

• Monitoring the time derivative of external magnetic field signals using induction sensors.
• Analyzing frequency spectra to capture changes in the machine's magnetic signature.
• Applying Shewhart control charts for anomaly detection in time series data.
• Utilizing correlation matrices to refine fault detection by filtering similar variation patterns.
• Developing a global change indicator for multivariable magnetic signature monitoring.

Main Results:

• The proposed method successfully detected stator and rotor faults in a laboratory setting.
• The global change indicator enabled automatic fault detection in a grid-synchronized generator.
• The methodology was validated on a 305 MVA hydroelectric power plant generator, detecting an incipient mechanical vibration fault.

Conclusions:

• The developed strategy provides an effective non-invasive approach for synchronous generator condition monitoring.
• The method offers a reliable and automated solution for detecting incipient and developing faults.
• This technique enhances the maintenance and operational safety of large-scale synchronous generators.