*3.2. FS-MPC Based on DSVM*

The approach that uses FS-MPC based on the DSVM strategy follows the same route for predicting the state variables that is used in the classical FS-MPC approach, as described in the previous subsection, except that the selected voltage vectors are obtained from various virtual voltage vectors for prefix time intervals [20]. These virtual voltage vectors are obtained by subdividing the space vector diagram (SVD) into M equal parts.

For example, Figure 3 shows the virtual voltage vectors where the space vector diagram is subdivided into three equal parts. The overall number of voltage vectors (T) in the space vector diagram is:

$$T(M) = 3 \cdot M \cdot (M+1) + 1.\tag{9}$$

According to Equation (9), the virtual voltage vectors, which are normally much higher, are calculated to obtain the current predictions and cost function. To enforce the actual output current vector to approach the reference current vector in the next step [20], the optimal voltage vector is applied to the inverter using the DSVM strategy during the entire time *Ts*. However, when taking into account all of the candidate voltage vectors for the current prediction, the computational burden increases dramatically, and the method becomes unsuitable for actual control systems.

**Figure 3.** Space vector diagram of conventional model predictive control with discrete space vector modulation (DSVM-MPC).
