Diagnosis of a Class of Faults in Distributed Parameter Systems

Abstract

In Distributed Parameter Systems (DPS), faults often appear as undesirable parametric variations in an unknown region of system space. In view of the difficulty of general identificata on of space-dependent parameters in nonlinear DPS and the distinctive features of fault diagnosis, a special fault identification technique is developed in this paper by the introduction of a Faulty Zone (FZ) concept. Fault diagnosis of DPS can thus be simplified as identification of three types of parameters: 1) the location of FZ; 2) the size of FZ and 3) the extent of parameter variations in FZ. This technique is applied successfully to the diagnosis of local diameter reduction of fluid transport pipeline by means of a hierarchic space division scheme and stochastic approximation algorithm. Simulation study and experimental results are included.