Parameter Identification of Hydraulic Manipulators Considering Physical Feasibility and Control Stability

Abstract

In the challenging applications of hydraulic robots, control performance in the full workspace is required for successful task completion. The model-based controller of the hydraulic manipulator is an effective means to improve control performance. However, most of the existing identification methods rely on least squares or weighted least squares, which may lead to the physically infeasible parameters. Although the physical feasibility of inertial parameters can be solved by the linear matrix inequality, physical feasibility and prior knowledge constraints of hydraulic parameters, such as bulk modulus and valve port coefficients, have not been considered in the identification process. Parameters that do not satisfy constraints will affect system stability in the model-based controller. In this paper, an identification framework is proposed, which integrates inertial, friction, hydraulic parameters physical feasibility. Compared with the non-identified parameters, the ratio of tracking error to maximum velocity with the proposed method is reduced by 50%-56%. Compared with the least squares identification, the control stability in the full workspace is guaranteed. The proposed method is applicable to serial hydraulic manipulators with arbitrary degrees of freedom and is supported by experimental analysis of a hydraulic manipulator.