Sensor fault diagnosis for urban sewer network management
Résumé
Considering the constraints of flood and pollution in urban sewer network, the control of hydraulic equipment now requires a higher reliability of measurements provided by different sensors. The latter falls with the field of fault diagnosis where the principle is generally based on consistency data checking between an observed behavior of the process provided by sensors and expected behavior provided by a model of the process. In this particular application, the problem of fault sensor diagnosis requires the availability of rainfall-runoff relationship in order to apply analytical redundancy-based diagnosis procedures. The multi-model is used to identify this relationship. The watershed model is then used to increase informational redundancy degree. Residual tolerances are derived from interval arithmetic. The proposed fault detection and isolation technique us successfully applied ion sensors implemented in the sewer network of Nancy, east of France.