On-line estimation of the mechanical stiffness of rod mechanisms

The stiffness of a mechanism is an important parameter for control design and position accuracy. A value of this parameter is usually obtained from a data-sheet of the coupling between the motor and the load, or from a measured frequency response. These methods are however not applicable to rod mechanisms, which are widely used in industry due to their rotary-linear conversion. Firstly, the load is typically directly connected to the rotor without a physical coupling. Secondly, the mechanical parameters and especially the inertia of these mechanisms are position and time dependent. Rod mechanisms are thus non-linear time-varying systems, which excludes off-line methods such as measuring a frequency response for estimation. This paper therefore presents a novel approach, based on the Sliding Discrete Fourier Transform (SDFT), for on-line estimation of the mechanical stiffness. The feasibility is verified with simulations on an equivalent two-mass system of an industrial pick and place unit, existing of a combined four-bar/slider-crank mechanism. Simulations show that accurate estimates are obtained and changing stiffness can be detected, which can serve as a new diagnostic tool