Supervision strategy of a solar volumetric receiver using NN and rule based techniques

Abstract

Most nonlinear processes suffer from lack of detectability when model based techniques are applied to IFDI (intelligent fault detection and isolation) tasks. Generally, all types of nonlinear processes will also suffer from lack of detectability due to the inherent ambiguity in discerning faults in the process, sensors and/or actuators. This work deals with a strategy to detect and isolate process and/or sensor faults by combining neural networks based on functional approximation procedures associated with recursive rule using techniques for a parity space approach. For this work, a case study dealing with the supervision of a solar volumetric receiver was performed using the proposed intelligent techniques, and produced reliable and acceptable IFDI results. © 2012 Springer-Verlag.