The use of dimensionality reduction techniques for fault detection and diagnosis in a AHU unit: critical assessment of its reliability

Hugo Geoffroy, Julien Berger, Benoît Colange, Sylvain Lespinats, Denys Dutykh

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

Fault detection and diagnosis (FDD) are important tools to perform on-going monitoring of the systems and help in their building commissioning. An innovative method is investigated based on combined data-driven and knowledge-based approaches. This article presents the method. In the first phase, a so-called operating map of the system is built using a dimension reduction method and numerical or experimental dataset. This map is composed of several regions corresponding to nominal operation and to specific faults. The second phase focuses on the FDD. The monitored data are projected on the map. According to the position, a clear and precise FDD can be carried. The method is applied to an air handling unit. The map is built using data generated with a building simulation programme. The reliability of the method is proven using experimental data of nominal and fault operation generated.

Original languageBritish English
JournalJournal of Building Performance Simulation
DOIs
StateAccepted/In press - 2022

Keywords

  • data-driven approach
  • dimensionality reduction technique
  • Fault detection and diagnosis
  • heating ventilation and air-conditioning systems
  • knowledge-based approach

Fingerprint

Dive into the research topics of 'The use of dimensionality reduction techniques for fault detection and diagnosis in a AHU unit: critical assessment of its reliability'. Together they form a unique fingerprint.

Cite this