A benchmark study approach to fault diagnosis of industrial process control systems

There have been several proposals and suggestions of benchmark studies for evaluating the performance of fault detection and isolation (FDI) methods allied to real industrial plant. The main aim of these benchmarks is to provide a training facility for the engineering community (both industry and academia) to gain an understanding and "feel" for the way in which various FDI methods can perform in a realistic control engineering application setting. This is deemed an essential step in the process of transferring the technology (often gained in the academic community) into real application. This presentation would provide the description and application of a benchmark scheme based on an intelligent electro-pneumatic valve actuator with in-situ (in the loop) testing and fault signalling in a sugar factory evaporisation process. The overall application would also be outlined and this involves on-line FDI and monitoring of the sensors and several actuators of a sugar juice evaporisation plant, providing overall monitoring of the plant under closed-loop control. The study was conducted within the Research Training Network "Development and Application of Methods for Actuator Diagnosis in Industrial Control Systems " DAMADICS {www.eng.hull.ac.uk/research/control}, funded by the European Commission in the Human Improvement Programme of Framework 5. The FDI benchmark is method-independent and based on an in-depth study of the phenomena that can lead to likely faults in valve actuator systems. The work to be presented uses a detailed consideration of the physical and electro-mechanical properties (and their modelling requirements) of an intelligent industrial actuator. The presentation would also include the typical engineering requirements of an actuator valve operating under challenging process conditions, together with the setting up of suitable performance indices for evaluating the FDI results. The results to be described correspond to real in the loop testing and FDI evaluation with injected fault signals.