Contact modeling and input-voltage-region based parametric identification for speed control of a standing wave linear ultrasonic motor

Abstract

Linear ultrasonic motor is a complicated electromechanical coupling system, precision speed control of which is a challenge for the researchers for difficulties in obtaining its accurate speed prediction model. The stator/slider contact mechanism, which can be described by a stick-slip-separation dynamics, greatly affects the output speed. Considering that practically the full physics of the system is often inaccessible to be modeled accurately for lack of prior knowledge of some physical/structural parameters, this paper presents a novel parametric contact model and the corresponding parametric identification procedure for steady-state speed prediction of a typical standing wave linear ultrasonic motor utilizing longitudinal and bending modes. Compared with the training data-driven general modeling, the approach retains physical insight into some important nonlinear characteristics in the contact process to improve prediction performance. The dynamic equations of the stator are deduced by using Timoshenko beam theory and Hamilton's principle and a new interface friction model including asperity contact based stick friction and normal ultrasonic oscillation related slip friction is developed. Besides, the distinctive feature of position-dependent nonlinearity of linear ultrasonic motor, which can be characterized by variation of the vibration of the stator at interface, is evaluated by introducing a variable named nominal axial force of the stator. In the study, this variable acts as the output of an embedded model in the contact model, parameters and structure of which are to be identified. Further, the effect of the region of input voltage on the contact behavior, is taken into account in the identification. In order to collect the input/output data for the identification procedure, an effective and synchronous data acquisition and transmission hardware/software experimental system is established. By employing off-line maximum likelihood estimation method, the input-voltage-region based parameters can be obtained. It is validated credibility of the speed prediction model under different loads and different input voltages with a prototype of linear ultrasonic motor actuated motion platform.