Validation de données débitmétriques issues de réseaux d'assainissement

Abstract : The objective of this thesis is to validate flow rate measurements collected from the sewer networks. Technical failures in the sensors or the existence of the disturbing phenomena produce anomalies to these measures. Our work consists in proposing a procedure of diagnosis off line able to detect and isolate the defects of measurements. This procedure is based on analytical models. We first established, a set of linear models used to generate faults indicators called residuals. These last ones allowed us to establish the experimental faults signatures matrix. Then, a first technique of diagnosis was considered. It consists in combining the test of Page for faults detection and on the calculation of Euclidean distance between the theoretical faults signatures and the experimental fault signature of each defect for the isolation of the faulty sensor. Two other techniques suggested, are based on the joint analysis of the generated residuals and the symptoms of the defects. One is called DMP (Model Diagnostic Processor) and the other is based on the fuzzy logic concepts. In the end, the performances of these techniques were calculated to be able to compare them. We improved the performances of the last considered technique by proposing a new operator of aggregation and by considering the persistence of the defects
Document type :
Theses
Complete list of metadatas

Cited literature [48 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-01752944
Contributor : Thèses Ul <>
Submitted on : Thursday, March 29, 2018 - 2:02:35 PM
Last modification on : Tuesday, April 10, 2018 - 1:27:50 AM

File

2007_HAMIOUD_F.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01752944, version 1

Collections

Citation

Farida Hamioud. Validation de données débitmétriques issues de réseaux d'assainissement. Autre. Institut National Polytechnique de Lorraine, 2007. Français. ⟨NNT : 2007INPL007N⟩. ⟨tel-01752944⟩

Share

Metrics

Record views

42

Files downloads

103