Consistency Index-Based Sensor Fault Detection System for Nuclear Power Plant Emergency Situations Using an LSTM...

Area of Science:

• Nuclear Engineering
• Sensor Technology
• Machine Learning

Background:

• Nuclear power plants (NPPs) have complex systems requiring reliable sensor monitoring.
• Existing sensor error detection methods are insufficient for emergency situations with drastic parameter changes.
• Reactor trips cause rapid shifts in plant parameters, challenging traditional monitoring techniques.

Purpose of the Study:

• To develop a machine learning model for detecting sensor errors specifically during NPP emergency situations.
• To introduce a novel 'consistency index' for assessing sensor reliability and measurement accuracy.
• To enable immediate and precise identification of malfunctioning sensors during critical events.

Main Methods:

• A machine learning model was developed using a consistency index to evaluate sensor data.
• The model was trained and tested on data from a compact nuclear simulator.
• Artificial sensor errors were injected into plant parameter data during simulated emergency scenarios.

Main Results:

• The trained machine learning system successfully distinguished between sensor error and error-free states.
• The consistency index effectively labeled the soundness of sensors based on measurement accuracy.
• The system demonstrated the capability to immediately detect sensor errors and pinpoint the faulty sensor.

Conclusions:

• The proposed machine learning model with a consistency index is effective for sensor error detection in NPPs during emergencies.
• This approach enhances the safety and reliability of nuclear power plant operations by ensuring sensor integrity.
• The method provides a reliable tool for real-time monitoring and fault diagnosis of sensors in critical nuclear environments.