Search

In this work a position controller for the permanent magnet synchronous motor is developed. The design is based on ideas of time-optimal control and predictive control and considers constraints for both torque and speed. The proposed controller takes the form of a state feedback. Its inputs are the position reference and the system state, and its output is the switching state to be applied by the power converter in the next sampling instant. The design procedure starts by solving a time-optimal control problem for a simple system: the triple integrator. The solution is then used to devise a state feedback rule for the motor. Although the control scheme does not have a completely cascaded structure, the approach taken establishes a hierarchy, with the error in the position having the highest priority. This separates the controller in two main building blocks, which work at different abstraction levels. The purpose of the first block is to generate references for the relevant states of the motor (speed and torque), based on long term predictions, which are realized using the triple integrator model. The second block is a receding horizon controller with a prediction horizon of one time step and a specially devised cost function, for the simultaneous control of speed, torque and stator currents. The performance of the controller is verified with experimental results.

This paper presents the design of a control scheme for the two-mass-system based on the Finite-Set Model Predictive Control method. The whole system state is controlled in a centralized way, so that the problem is solved more holistically, that is to say, using all degrees of freedom available. Constraints in the prediction horizon, given by the limited computation power and speed of the system dynamics, and the consequent myopic perspective of the controller, are solved by means of an ad-hoc cost function. The control is done using measurements of the stator currents and the speed of the driving machine. The rest of the system state is estimated using a Kalman Filter.

In this paper the use of a Reduced Order Extended Kalman Filter, to estimate the rotor position of a Permanent Magnet Synchronous Motor, in coordination with a Predictive Torque Controller is proposed. The observer integrates the information provided by the fundamental model and the anisotropies of the machine, allowing for a wide operation range. The scheme takes advantage of the nature of the predictive controller, which applies one switching state of the inverter at a time for a whole sampling period, eliminating problems associated with pulse width modulation at low modulation indexes.

Multilevel converters can reach medium-voltage operation, increasing the efficiency in high power applications. Among the existing multilevel converter topologies, modular multilevel converter provides the advantages of high modularity, availability and high power quality. Moreover, the main advantage compared to cascaded multilevel converters is the absence of input transformer; reducing the size, cost and cooling requirements. When used as AC-AC converter, it usually requires a sophisticated controller to decouple the input and output frequency components. In this paper a predictive control approach is proposed to control a single-phase AC-AC modular multilevel converter. The controller minimizes the currents error and keeps the DC voltage balancing featuring a simple implementation, high input power factor and precise control of the output current. Moreover, the DC voltages are kept at the reference value. Simulation results showing the performance of the predictive control scheme are presented.

This paper presents a predictive strategy for the control of a permanent magnet synchronous motor (PMSM). Our approach allows one to simultaneously manipulate speed and electrical variables, without using any auxiliary linear controller. The high sampling rate, characteristic of this technique, and the limited precision of the typical angle measurement devices, present serious difficulties for this technique to work in practice. These issues are dealt with in the paper and a solution based on an extended Kalman filter is proposed. The effectiveness of the proposed approach is evaluated through simulations.