Design of an optimal observer for making liquid level control loop robust to variations in transmission parameters. Issue 1 (1st January 2020)

• Record Type: Journal Article
• Authors: Krishnan Venkata, Santhosh; E Silva, Nathan Noronha; Nair, Nanditha
• Editors: Tan, Yuegang
• Is Part Of: Cogent engineering. Volume 7:Issue 1(2020)
• Publication Date: 2020-01-01
• Subjects: Kalman filter -- Luenberger -- networked control system -- observer -- quantizer delay -- transmission delay; Engineering -- Periodicals; Technology -- Periodicals; Engineering; Technology; Periodicals; 620
• DOI: 10.1080/23311916.2020.1840688
• Languages: English
• ISSNs: 2331-1916

Abstract

Abstract: The interconnected system has become the need of the hour in the present day industrial automation. Interconnection is established by conventional wired or by wireless techniques. In either case for the ease of data transmission discretization is essential as process variables are analog. This paper discusses the design of an observer to estimate the effects of Data Acquisition Cards (DAQ) like transmission delay and quantizer delay on a networked liquid level control loop. The objective of the proposed work is to accurately control the level of liquid in a tank, even if there are variations in the performance of the data acquisition card used to transmit the data in between the actual plant and computer. This can be achieved by designing an observer that will estimate the effect of data acquisition card parameters like transmission delay and quantizer delay on the system behavior of the process. A difference of observer output and existing process which is affected by the transmission parameter is filtered, so as eliminate the effect of variation in data acquisition parameters on the process. In the proposed work observer is designed using techniques like Luenberger and Kalman filter approach. Performance analysis shows that a Kalman filter-based observer produces better results as compared to a Luenberger observer. Results show that a Kalman filter-based observer produced a root mean square error of 0.015, and the root mean square of percentage overshoot of 1.16 for the test with the practical setup.