For many position controlled actuators it is desirable to achieve the demanded position in the minimum possible time within the limitations imposed by available hardware. This can be achieved by means of the 'time optimal control', which is a form of 'bang-bang' control, in which control variable switches between its saturation limits. Such control system then suffers from limit cycling with no damping upon approaching the reference position. To solve this problem some modifications of the 'time optimal control' were introduced. Small price to pay for the limit cycle elimination measures and the assumption of zero friction is that the settling time is increased slightly with respect to the theoretical one and therefore the phrase 'near time optimal'is introduced. For the development of suitable control algorithm for changing the position of one degree of freedom the principles vector control and time optimal control are combined to form 'near time optimal' algorithm. Control chatter is eliminated from the model by introduction of a boundary layer and a linear term to the switching boundary equation. The system automatically counteracts load torque by producing a nearly equal and opposite control torque component, provided by a load torque observer. The simulations and preliminary experimental results presented show good agreement with theoretical predictions.
展开▼
