Robust state estimation of the anaerobic digestion process for municipal organic waste using an unscented Kalman filter. (January 2023)

• Record Type: Journal Article
• Authors: Raeyatdoost, Niloofar; Bongards, Michael; Bäck, Thomas; Wolf, Christian
• Is Part Of: Journal of process control. Volume 121(2023)
• Page Start: 50
• Page End: 59
• Publication Date: 2023-01
• Subjects: State estimation -- H-infinity unscented Kalman filter -- Robustness -- Anaerobic digestion -- Model parameter estimation -- Organic fraction of municipal solid waste; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2022.11.013
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: The stable and sustainable operation of anaerobic digestion (AD 1 ) plants is crucial for their feasible long-term operation. Due to the high non-linearity of the AD process, the main contribution of this paper is to investigate the performance of a robust unscented Kalman filter (UKF) for dynamic process state estimation for the organic fraction of municipal solid waste (OFMSW 2 ). A traditional UKF performance degrades in case of dynamical model uncertainties or the presence of unknown Gaussian and non-Gaussian noises. The utilized robust UKF proposed by Zhao and Mili (2019) addresses this issue by improving the accuracy of state estimation using the H-infinity filtering theory. The performance of the utilized estimator is tested for the case of uncertainties within the AD model and the substrate feed, as well as unknown Gaussian and non-Gaussian process and measurement noise. Results show that the developed robust UKF estimator achieves higher accuracy and presents higher robustness against uncertainties. Besides, the mathematical model considered for the simulation and the state estimator both rely on the extended version of the AD model No. 2 (E-AM2 3 ) including a hydrolysis step. This model is a simplified AD model, which is a suitable choice for control and estimation purposes providing a reasonable trade-off between complexity and accuracy. In this paper, the developed E-AM2 model is adapted for the AD process of OFMSW. Highlights: Adaptation of a simple dynamic model for anaerobic digestion (AD) of organic waste. Development of a robust state estimation method using the adapted AD model. Performance of the developed state estimator in presence of uncertainties.