A falsification perspective on model reference adaptive control

Section: Introductionmentioning

confidence: 99%

Force control of a reciprocating surface grinder using unfalsification and learning concept

Razavi

2001

“…Let us recall, that the falsification process and the standard MRAC hypothesis [16] were used in [1]. In this work, however, we advanced to deal with the cases where adaptive control was most needed, where those hypothesis did not hold, as stated in [16].…”

Section: Discussionmentioning

confidence: 99%

“…We first dealt with the model reference adaptive control problem in Reference [1]. In that work, we solved the problem, with its standard hypothesis, using a falsification perspective, showing that perfect tracking could be achieved in finite time and that the solution was given by one point, which could be computed in one step.…”

Section: Our Previous Work On the Model Reference Adaptive Control Prmentioning

confidence: 99%

“…The unfalsified control paradigm rules out controllers from a given class of candidates based on the check of consistency between data, performance specification and control laws [1,[7][8][9][10][11][12][13][14]. The performance specification has often been given in terms of a solution to an expanding collection of quadratic inequalities [9,10], with new quadratic inequalities being continually added as time increases and measurement data accumulates.…”

Section: Unfalsified Controlmentioning

confidence: 99%

“…In order to define the class of candidate controllers, we first define a vector of filters as in [1,2]. Notice that we define a vector of filters, and not just a vector of time-domain filtered signals as in [16,17].…”

Section: The Class Of Candidate Controllersmentioning

confidence: 99%

See 2 more Smart Citations

Unfalsified model reference adaptive control using the ellipsoid algorithm

Cabral

Safonov

2004

The unfalsified control paradigm which uses as plant information, experimental trajectories of plant signals, is applied to the model reference adaptive control problem. The use of an ellipsoid algorithm overcomes the need of gridding the parameter space used in prior applications of unfalsified control. The application of the ellipsoid algorithm produces a sequence of decreasing volume ellipsoids which contains the set of unfalsified candidates. Simulation results are provided for the control of an unstable plant.

Unfalsified direct adaptive control of a two‐link robot arm

Tsao¹,

Safonov

2001

An application of unfalsi"ed control theory to the design of a switching adaptive controller for a non-linear robot manipulator is described. In the unfalsi"ed control approach, candidate controllers are eliminated and discarded when their ability to meet performance goals is falsi"ed by evolving experimental data. Switching occurs when the currently active control law is among those falsi"ed. In this design study, the candidate controllers are non-linear, and have a non-linear &computed torque' control structure with four switchable parameters corresponding to unknown masses, inertias and other dynamical coe$cients of a class of ideal, but imperfect robot arm models. Simulations con"rm that our unfalsi"ed switching controller permits signi"cantly more precise and rapid parameter adjustments than a conventional adaptation law having continuous parameter update rules, especially when the manipulator arm is subject to sudden random changes in mass or load properties.