Acceleration profile is a little related to feedback loop. I have worked with state feedback control instead of PID for a while and it works great although adding integrator have puzzled me a little bit.
Ordinary integrator works between set reference value and measured value and this will only work for a dead beat controller. Then gain is lower it will integrate not only the current model and measurement errors but also the part because real have not yet had enough time to change. Then integrating the model error which is the difference between change predicted by model and measured value integrator seems to work much better. In theories for state feedback control they use the zero point and claim it could easily be expanded to any point. All articles and books I have read about state feedback including linear quadratic control with or without noise (LQR, LQG) they use difference between set reference value for position/angle and measured position/angle. This must be a little bit wrong and it should be the differences for each state because as is now all differences except one is from zero. It should be: (angle_ref - angle_mesaured)*gain1 - (velocity_ref - velocity_measured)*gain2 - (current_ref - current_measured)*gain3. From the set reference value for position the model is used used to calculate set reference value for all three values. There may also be an individual integrator for each. Nicklas Karlsson ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
