ONLINE FAULT-DETECTION OF FLOW-INJECTION ANALYSIS SYSTEMS BASED ON RECURSIVE PARAMETER-ESTIMATION

Effective automated supervision can help to ensure the reliable operat ion of complex flow-injection analysis (FIA) systems. As an important element of a supervisory system, fast fault detection of the FIA syste ms is required. In this paper, a model-based fault detection method ba sed on the identification of the model parameters is developed. The fa ult detection system consists of the three levels estimation, filterin g and evaluation of the model parameters. The model order and the time delay of the system are determined on-line. The recursive fixed memor y (RFM) method is used to estimate model parameters. The fault detecti on of a FIA system is performed by means of filtering the estimated mo del parameters through a high- and low-pass filter for separation of f aults with different magnitude in their dynamics. The application to d ifferent practical examples confirms that the newly developed method o ffers a very effective way for fault detection in the FIA process.