A novel adaptive-gain technique for high-order sliding-mode observers with application to electro-hydraulic systems. (October 2020)

• Record Type: Journal Article
• Authors: Palli, Gianluca; Strano, Salvatore; Terzo, Mario
• Is Part Of: Mechanical systems and signal processing. Volume 144(2020)
• Publication Date: 2020-10
• Subjects: Hydraulic actuators -- High-order observers -- Adaptive gains -- Nonlinear systems; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2020.106875
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A novel technique for adaptive gain selection in high-order sliding mode observers. State and disturbance estimation on a hydraulic actuator is experimentally evaluated. Comparison with fixed gains and a classic gain adaptation technique is made. The performance are comparable or superior to other gain selection methods. Better gain convergence rate in case of large estimation errors and noisy measures. Abstract: In this paper, a novel technique for adaptive gain selection in high-order sliding mode observers to estimate the state and the disturbance acting on a hydraulic actuators is designed and tested. The proposed gain adaptation technique is based on the evaluation of the absolute value of the errors computed between the real measurements available on the system and their estimation provided by the observer. Experimental results are reported to validate the effectiveness of the proposed technique in a real context. The experiment are conducted in two different conditions, i.e. without external load and with an unknown visco-elastic seismic isolator attached to the hydraulic actuator. The proposed gain-adaptation method is also compared with the case of manually-tuned gains and a classic gain adaptation technique reported in literature, in which the gain adaptation is based on fixed variation rates. The proposed method provides performance comparable or superior to both manual and classic adaptive gain selection. Moreover, with respect to classic techniques, the proposed one has the advantage of improving the convergence rate in case of large estimation errors and limiting the growing of observer gains in case of noisy measurements.