A Fault Detection System for a Geothermal Heat Exchanger Sensor Based on Intelligent Techniques

Area of Science:

• Engineering
• Machine Learning
• Sensor Technology

Background:

• Real-world engineering systems often face sensor malfunctions, impacting operational reliability.
• Accurate sensor data is critical for effective fault detection and recovery processes.

Purpose of the Study:

• To develop a robust methodology for identifying malfunctioning sensors.
• To establish a strategy for determining correct measurement values when sensor failures occur.

Main Methods:

• Utilized six machine learning techniques: Random decision forests, gradient boosting, extremely randomized trees, adaptive boosting, k-nearest neighbors, and shallow neural networks.
• Modeled sensor behavior using data from a geothermal heat exchanger in a bioclimatic house.

Main Results:

• The proposed methodology demonstrated high effectiveness in identifying sensor faults.
• The approach successfully established correct measurement values, validating its practical utility.

Conclusions:

• The developed methodology offers a satisfactory solution for sensor fault detection and recovery in practical engineering systems.
• Machine learning techniques provide a powerful tool for enhancing the reliability of sensor networks.