There is a nonlinear dynamic problem in the response of multi-dimension force/torque sensors. The Hammerstein model is introduced to describe that problem, and on this basis, an algorithm named "one-stage identification" is presented. The linear dynamic subsystem and nonlinear static subsystem of Hammerstein model are obtained after the identification, thus the nonlinear dynamic system is affirmed. Based on the "one-stage identification", a "two-step compensation" method is put forward, which consists of a linear dynamic compensation subsystem and a nonlinear static correction subsystem. Results of simulation and experimental data processing show that the "one-stage identification" algorithm has a rather high accuracy and the "two-step compensation" method improves the dynamic response performance of sensors. And then, on the basis of ADSP2189M EZ-KIT Lite Evaluation Board, a nonlinear dynamic compensation system is developed for sensors, which not only can produce the nonlinear dynamic responses, but also can realize the nonlinear dynamic compensation in real time. Otherwise, some problems of the dynamic calibration of multi-dimension force/torque sensors are studied, and the author of this thesis also gives out his own point of view.
|