Nowadays, the need of welding industry's to improve weld quality has led to the consideration of robotic welding. The use of articulated industrial robots for welding has many challenges. Because some robots do not have the capability of online error compensating of the seam track. Therefore, in order to remove the welding seam tracking error, the use of an auxiliary mechanism is proposed in this article. This mechanism is a table with 1-degree of freedom (dof), which produces a continuous motion in workpiece under the welding torch. The rotational motion of the motor is transformed into a translational motion of the workpiece by a ball-screw system, where this linear motion compensates the tracking error. Since in the welding process, relative motion accuracy of the workpiece and the welding torch is crucial, proper control of the interface table ensures the weld quality. In this paper, two different methods for controlling the table with 1-dof are studied. In the first method, due to the complexity of friction model of the ball-screw mechanism and the presence of nonlinear terms, this part of the model is considered as an external disturbance, and, then, a PID controller for the linear part is designed. In the second method, known as feedback linearization, a control law is designed for that the tracking error tends to zero by passing time. Throught a comparison between the simulation results, the second control method demonstrates better precision relating the first controller. While the error of PID controller equals to 3 mm and the second controller’s error does not go beyond 0.5 mm. At last, the experimental cell used for the robotic welding is introduced to evaluate the mentioned results.

Piezoelectric bending actuators have been extensively utilized in recent years. Two major modeling methods, lumped and continuous, have been generally proposed in previous researches for these actuators. The lumped method can only express the transverse vibration of one specified point on the actuator. In addition, the effect of higher vibrational modes has been ignored. Hence, continuous dynamic models have been proposed to rectify the mentioned drawbacks. In this method, linear constitutive equations for low voltage applications are usually applied. But, the main challenge in continuous modeling of piezoelectric actuators is the hysteresis nonlinear phenomenon caused by excitation voltages. In this paper, piezoelectric nonlinear constitutive equations have been employed to carry out the continuous dynamic model for two general types of bending actuators i.e. Series and Parallel. In addition, zero dynamic analysis for nonlinear systems has been applied to clarify the effect of higher vibrational modes the actuator dynamic behavior based on the location of Experimental results show the maximum error 1.44 and 1.2% in the identification of first and second modes, respectively, and the maximum error 2.89% in the modeling of actuator nonlinear behavior by two modes. These results validate the efficiency of the proposed dynamic model to express the actuator nonlinear behavior, dynamic analysis, and its superiority over conventional models with one mode.

Today, piezoelectric actuators are widely used in micro-positioning applications due to unique features such as high precision, fast response and high natural frequency. Despite the aforementioned characteristics, nonlinear characteristics such as hysteresis deteriorate the precision of piezoelectric actuators. In order to reduce the effect of hysteresis in control applications, external sensors are used for feedback control schemes. But, high costs and space limitations are prohibitive factors which limit the application of external sensors. Hence, an alternative is using self-sensing methods that is based on electromechanical characteristics of piezoelectric materials which eventually eliminate external sensors. In this research, self-sensing method is applied for position estimation in piezoelectric actuators. The most conventional method is based on the linear relation of electrical charge and actuator position which the position can be estimated by measuring the actuator charge. But this method is faced with serious challenges due to charge drift, especially at low frequencies. For this purpose, a method for modeling and compensating of charge drift is proposed. Then, by linearization of the electric charge-position relation, the self-sensing method is implemented based on the compensated electric charge measurement. Experiments have confirmed that this method can effectively estimate the actuator position with 1.5% estimation error in the presence of charge leakage.