Finite set model predictive control of the PMSM with sine-wave filter

In the frame of this work a predictive control for a permanent magnet synchronous machine with LC-Filter aiming sinusoidal voltage at the machine terminals is presented. The proposed scheme assumes a virtual multilevel inverter for the optimization of the cost function and utilizes a conventional pulse width modulator (PWM) thus being a finite control set predictive control scheme. A first estimation of the voltage space phasor, that has to be synthesized by the inverter, is computed by using the equations of the mathematical model of the system enhanced by a Luenberger observer and based on measured inverter currents and the shaft position. The switching patterns are calculated by assuming that the "virtual" inverter has an arbitrary high number of levels. The cost function in a predictive control scheme is minimized by defining a mesh of voltage space phasors that corresponds to the output states of the “virtual” multilevel inverter. The center of this mesh with a finite number of elements corresponds to the first calculated value of the reference voltage. For the minimization only the points belonging to the mesh are considered and the voltage space phasor chosen in this way is then used for the calculation of the switching signals in a conventional space phasor modulator and sent to the real two-level inverter. The proposed scheme results in an easy implementation and delivers good dynamics in the torque behavior. To ensure the convergence on the digital implementation, the Runge-Kutta method is used in each integration step. The proposed predictive controller is complemented by means of sensorless control as well as on-line identification and adaptive parameter tuning, both based on the voltage model method of the machine and working at the same time. This method identifies and adapts all machine parameters simultaneously (resistance and flux) and it is applicable to high-dynamic drives.