On-line fault detection of flow-injection analysis systems based on recursive parameter estimation

Effective automated supervision can help to ensure the reliable operation of complex flow-injection analysis (FIA) systems. As an important element of a supervisory system, fast fault detection of the FIA systems is required. In this paper, a model-based fault detection method based on the identification of the model parameters is developed. The fault detection system consists of the three levels estimation, filtering and evaluation of the model parameters. The model order and the time delay of the system are determined on-line. The recursive fixed memory (RFM) method is used to estimate model parameters. The fault detection of a FIA system is performed by means of filtering the estimated model parameters through a high- and low-pass filter for separation of faults with different magnitude in their dynamics. The application to different practical examples confirms that the newly developed method offers a very effective way for fault detection in the FIA process.