Skip to Main content Skip to Navigation

Diagnosis in building : new challenges

Abstract : Fault diagnosis and maintenance of a whole-building system is a complex task to perform. Available building fault detection and diagnosis tools are only capableof performing fault detection using behavioral constraints analysis. The thesis of Mahendra Pratap Singh proposes to use heterogeneous tests with validity constraints in the context of building fault diagnosis but the proposed approach assumes that the sensors are reliable. Nevertheless, validity constraints are checked with potentially faulty sensors. If these sensors are faulty, the diagnostic result is not guarantee and there is a need for method to prove the test as well as global diagnoses.To make a test, data are required from different parts: meteorological, human and physical parts. However, the data gaps is the main sensor fault in buildings. Sensor values are not uniformly sampled and there is a need to decide from which delay the sensor becomes faulty?The objective of this work is to highlight these challenges as well as to provide a strategy about how to solve them. Three solutions for diagnosis in building are proposed1-A level of completeness for better formalizing validity.In this work, we make the hypothesis that there is no precise global model for a building system but there is contextual models with limited validity. The validity is measured with potentially faulty sensors. The completeness level is proposed as a method to prove if a test space is fully covered or not i.e to assess the level of validity of a test.2-A confidence level for proving global diagnosis.A test is characterized by thresholds i.e the behavioral constraint is either satisfied or unsatisfied. Uncertainty is related to the validity constraints. Indeed, it is difficult to set a threshold for the level of completeness from which one can say that a test is valid.Diagnostic results are calculated from a set of tests, each one defined by itscompleteness level. The contribution is to propose a solution to compute the confidence level of a global diagnosis deduced from a set of tests whose some of them have a completeness level lower than 1. A method based on fuzzy logic reasoning is used for this purpose.3- Automatic thresholding for sensor data gap detection.The delay depends on the measured value and the type of sensor. The objective is toidentify from which delay a sensor become faulty. Two techniques are proposed: a time series analysis and a statistical approaches.Different applications have been studied for validation: an office at G-SCOP lab,an appartement at Grenoble and a platform in the University of southern Denmark.
Document type :
Complete list of metadatas

Cited literature [140 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Thursday, September 10, 2020 - 11:44:33 AM
Last modification on : Wednesday, October 14, 2020 - 4:19:12 AM


Version validated by the jury (STAR)


  • HAL Id : tel-02935337, version 1



Houda Najeh. Diagnosis in building : new challenges. Automatic. Université Grenoble Alpes; École nationale d'ingénieurs de Gabès (Tunisie), 2019. English. ⟨NNT : 2019GREAT110⟩. ⟨tel-02935337⟩



Record views


Files downloads