An experimental study on the direct & indirect dynamic identification of an over-constrained 3-DOF decoupled parallel mechanism

This paper provides the state of the art of obtaining the dynamic model of an over-constrained 3-DOF translational parallel mechanism, namely, Tripteron, by performing system identification concept in which friction is taken into account. To this end, as the first step, in order to circumvent the problem of being over-constrained, i.e., inconsistent numbers of equations and unknown parameters in the dynamic model, some modifications on the kinematic arrangement of the constituted limbs are applied, in such a way that preserve the degrees-of-freedom performed by the mechanism under study, the so-called Tripteron with three translational degrees-of-freedom. Then, in order to have a close-to-reality dynamic model, the model is combined with the model of the linear module friction. Afterward, by using the practical data, the optimum values for the uncertain parameters of the inverse dynamic and friction models are obtained by a white-box identification approach based on Genetic algorithm. In addition, in order to propose a less complex model, a model based on the polynomial regressors of position, velocity, acceleration and applied torque of each axis is proposed. Finally, by proposing some criterion, the paper concludes by a comprehensive discussion on the performance of the obtained models.