Development of online data filtering based on Kalman filter

Abstract

Knowledge of accurate process measurements in the form of Flow, temperature and pressure strongly affect product quality, process real time optimization and control, plant safety and plant profitability. The paper reports an experience with online data filtering in Naphtha Hydrotreater setup. First, pilot plant data is analyzed for detecting and removing faulty data and gross errors. To remove noise hidden in the process data, a fast and adaptive data denoising technique is proposed. The proposed technique is based on the recursive least square to identify the pilot plant model and the Kalman filter to reconcile noisy data. This technique offers competitive advantages over conventional approaches: Independent and adaptive model and less computation time. From several pilot runs, the proposed technique has shown good performance in terms of accuracy and speed.

DOI: http://dx.doi.org/10.3329/cerb.v17i1.22913

Chemical Engineering Research Bulletin 17(2015) 11-17