1 Introduction

In the recent years, electric vehicles (EVs) received a lot of consideration as they contribute to the solution of the environmental problems in two phases. Firstly, the oil resources are limited, and hence, the enlarged use of EVs conserves oil resources for critical applications. Secondly, the use of EVs reduces the use of internal combustion engines, and hence, the oil emissions are reduced [1]. Electric motors are the muscles of EVs that convert electric energy to mechanical energy that drives the vehicle. The drive electric motor must grant improved torque and power density, extended torque-speed curve, high efficiency, and feasible cost. Induction motors are employed in the Tesla Model S to avoid the usage of expensive rare-earth magnets and to benefit from the high reliability; however, the high starting current represents a challenge for the inverter and battery design [2]. Interior permanent magnet (IPM) machines provide high efficiency and high dynamic performance. They are employed in the Nissan Leaf, Toyota Prius 2004, Camry 2007, and Lexus 2005 [3].

The two-level voltage source inverters (VSIs) are the classical form of power converters in IPM motor drives. However, nowadays, multilevel inverters attract great attention thanks to their merits such as high-power quality, reduced switching loss, low \(\mathrm{d}v/\mathrm{d}t\) stress, low common-mode voltages, and reduced filter size [4]. These advantages can be met by the T-type multilevel inverter [5,6,7], which is considered as a perfect choice for high-power medium-voltage IPM drives.

Generally, the characteristic operation of IPMs can be divided into three regions, namely the constant torque region, flux-weakening (FW) (or constant power) region, and infinite-speed region [8]. During the constant torque region, a maximum torque per ampere (MTPA) strategy is essential to improve the motor efficiency, whereas an FW and a maximum torque per volt (MTPV) technology are required in the FW and infinite-speed regions, respectively, to extend the operating speed with the same voltage constraints [9]. These schemes can be simply implemented offline by using lookup tables [10], analytical models [11], or approximated equations [12, 13] for reference currents generation. The online MTPA/FW methods can be realized by using parameters estimation [14], signal injection [15], or optimization techniques [16, 17]. The online tracking of the MTPA/FW schemes is more accurate than offline counterpart [18]. However, online MTPA/FW schemes have excessive computational processes, which increases the hardware requirements [19]. Therefore, a compromise between tracking accuracy and hardware requirements is necessary for efficient MTPA/FW control of IPMs.

Due to system and controller complexity, the performance potentials of T-type MLI-fed IPM in the FW region have not been fully exploited. The performance of a permanent magnet synchronous machines (PMSM) fed by a T-type MLI has been investigated by [20, 21], where the power losses, efficiency and dynamic response are analysed. However, this study is limited to the constant torque region, and a classical proportional-integral (PI) controller with a carrier-based PWM technique is employed. The fault-tolerant capability and the efficiency of low-voltage low-power applications are investigated in [7, 22], respectively.

The reference currents of the MTPA/FW trajectories should be tracked by using a proper control approach, which is characterized by fast tracking response, robust operation, and constraints and nonlinear controllability. Despite the existence of simple current control schemes for IPMs such as PI or hysteresis-based controllers, only model predictive control (MPC) can satisfy the required characteristics [23]. A simple and classical form of MPC is finite control set (FCS)-MPC, which is commonly used for power electronics and motor drives [24]. However, the calculation load of FCS-MPC is the main challenging for their application in IPM drives, especially when it is applied to a T-type multi-level-based drive with multiple objectives and many switching states [25]. In addition, the attached MTPA/FW calculations further increase the computational load. Focusing on power loss reduction, a low-complexity MPC scheme is applied on an improved T-type MLI topology [26], where the dc-link voltage balance is implemented by control-set vectors grouping. Although the switching frequency of FCS-MPC is naturally variable, a constant switching frequency without the need for tuning the weighting factors is presented in [27], however, it is proposed for single-phase T-type inverters. A multiport magnetic-link T-type single-phase inverter is presented in [28]. Still, the integration of MPC with T-type MLI-fed IPM is not a common solution due to the computational load.

In this paper, a T-type multilevel inverter is employed to drive an IPM motor drive. The IPM drive is set to operate in the MTPA, FW, and infinite-speed regions. The optimal current trajectories in the three operation regions are provided by a feedforward calculation based on IPM model analysis. Furthermore, the voltage constraint is considered by using a PI-based voltage feedback regulator, which assures safe and efficient operation of the IPM within the voltage limits. The proposed scheme has the potential of simple structure compared with existing FW techniques. To avoid problems of using cascaded PI controllers for the speed and current control, a simplified FCS-MPC control is employed, and hence, the limitations that arise due to using the SV-PWM is minimized. The integration of MPC with simplified MTPA/FW controller opens new research horizons for fully implemented FW schemes with low-complexity MPC techniques. The system performance is validated by extensive simulation scenarios and real-time simulations.

The paper is organized as follows: Sect. 2 presents the operation and mathematical modelling of the IPM drives fed by T-types MLI. Section 3 exhibits the proposed FW control. The simulation results are discussed in Sect. 4. The real-time simulation is exhibited in Sect. 5. Finally, the conclusions are summarized in Sect. 6.

2 The operation of the proposed drive

2.1 T-type multilevel inverter

The 3L T-type MLI is constructed by connecting two converter circuits; the first is the conventional three-phase VSI and the second is the auxiliary circuit, see Fig. 1. The VSI operates as a full-bridge two-level inverter, namely the output pole voltage \({{V}_{\mathrm{out}}=+V}_{\mathrm{dc}}/2\) and \({{V}_{\mathrm{out}}=-V}_{\mathrm{dc}}/2\). The auxiliary circuit is used to clamp the output voltage to the midpoint \(O\); hence, the third level,\({ V}_{\mathrm{out}}=zero\), is introduced in the output voltage waveform. The two direction switches (\({S}_{2x}\) and \({S}_{3x}\)) with anti-parallel diodes are designed to enable current flow in the two directions. Table 1 presents the switching states for one phase output voltage.

Fig. 1
figure 1

T-type MLI fed an IPM machine

Table 1 Possible switching states for one phase (\(x=a, b, or c\))

For a three-phase 3L MLI, 27 states are feasible, as presented in the \(\alpha -\beta \) plane in Fig. 2. The voltage vectors \({V}_{1}\) to \({V}_{6}\) represent the six large vectors of the conventional hexagon as obtained by the three-phase VSI. Vectors \({V}_{7}\) to \({V}_{12}\) represent the medium six vectors. Vectors \({V}_{13}\) to \({V}_{18}\) represent six small vectors with twelve redundant switching states, six of them are positive vectors (P-type) and the other six states are negative (N-type). Finally, three zero vectors (\({Z}_{1}\), \({Z}_{2}\), and \({Z}_{3}\)) are located in the origin. The 27 vectors with switching states can be classified as shown in Table 2.

Fig. 2
figure 2

Possible switching vectors in the \(\alpha -\beta \) plane

Table 2 Classification of switching vectors [29]

2.2 IPM machine operation

The mathematical model of the IPM machine in the synchronous reference frame can be expressed as:

$${v}_{\mathrm{ds}}={R}_{\text{s}}{i}_{\text{ds}}-{\omega }_{\text{e}}{L}_{\text{q}}{i}_{\text{qs}}+{L}_{\text{d}}\frac{d{i}_{\text{ds}}}{dt}$$
(1)
$${v}_{\text{qs}}={R}_{\text{s}}{i}_{\text{qs}}+{\omega }_{\text{e}}({L}_{\text{d}}{i}_{\text{ds}}+{\lambda }_{\text{PM}})+{L}_{\text{q}}\frac{d{i}_{\text{qs}}}{dt}$$
(2)
$${T}_{\text{d}}=\frac{3}{2}{N}_{P}[{\lambda }_{\text{PM}}{i}_{\text{qs}}+({L}_{\text{d}}-{L}_{\text{q}}){i}_{\text{qs}}{i}_{\text{ds}}]$$
(3)

where vds and vqs are the stator d- and q-axis voltages, respectively. ids and iqs are the stator d- and q-axis currents, respectively. Rs, ωe, Lq, Ld, and λPM are the stator resistance, the electrical angular speed, q-axis inductance, d-axis inductance, and the PM flux-linkage, respectively. Td and \({N}_{P}\) are the developed torque and the number of pole pairs, respectively.

To ensure the optimal utilization of the inverter and machine, the current and voltage constraints in (4) and (5) must be fulfilled.

$${I}_{\text{s}}= \sqrt{{i}_{\text{sd}}^{2}+{i}_{\text{sq}}^{2}}\le {I}_{\text{sMax}}$$
(4)
$${V}_{\text{s}}= \sqrt{{v}_{\text{sd}}^{2}+{v}_{\text{sq}}^{2}}\le {V}_{\text{sMax}}$$
(5)

where IsMax is the maximum line current. VsMax is the maximum supply voltage based on the dc-bus voltage, Vdc, and the applied modulation strategy as can be calculated by (6).

$${V}_{\text{sMax}}= \left\{\begin{array}{c} \frac{1}{2}{V}_{\text{dc}}, sine-PWM \\ { \frac{1}{\sqrt{3}}V}_{\text{dc}}, space-vector PWM\\ \frac{2}{\pi }{V}_{\text{dc}}, six-step operation\end{array}\right.$$
(6)

The current and voltage limits imposed by (4) and (5) force the drive to operate within a predefined profile for optimal efficiency and dynamic operation of the IPM drive. As can be seen from Fig. 3, the MTPA operation is accomplished when the drive current follows the \(oA\) trajectory, where the operating point is defined by the intersection point with the torque hyperbola. Point \(A\) defines the rated voltage, current, torque, and speed of the drive. When the drive speed exceeds the rated speed, the current vector moves on the \(AB\) trajectory. The current vector preserves the maximum current value, while the d-axis current is employed to weaken the flux-linkage, i.e. the drive operates in the flux-weakening region. To preserve the maximum limit of current, the q-axis component is decreased along with the developed torque. For IPM machines, it is possible to design the characteristic current point, \(C=\frac{{\lambda }_{\text{PM}}}{{L}_{\text{d}}}\), inside the current limit circle, then the MTPV region exists, and the drive can theoretically operate at infinite speed. This mode of operation is defined by a current vector moves on the \(BC\) trajectory, i.e. decreasing both the d-q axis current components. This leads to less flux-linkage weakening and extra torque reduction.

Fig. 3
figure 3

IPM drive operating trajectories in the \({i}_{\text{d}}-{i}_{\text{q}}\) frame

Based on this analysis the control algorithm is defined as follows:

  • For rotor speed below the rated speed, the current vector is adjusted as (7) and (8).

    $${i}_{\text{dMTPA}}=\frac{{\lambda }_{\text{PM}}}{2({L}_{\text{q}}-{L}_{\text{d}})}-\sqrt{\frac{{\lambda }_{\text{PM}}^{2}}{{4\left({L}_{\text{q}}-{L}_{\text{d}}\right)}^{2}}+{\left({i}_{\text{q}}^{*}\right)}^{2}}$$
    (7)
    $${i}_{qMTPA}={i}_{\text{q}}^{*}={k}_{p}\left({\omega }_{m}^{*}-{\omega }_{m}\right)+{k}_{i}{\int }_{0}^{{T}_{\text{s}}}\left({\omega }_{m}^{*}-{\omega }_{m}\right)dt \le \sqrt{{I}_{\text{sMax}}^{2}-{i}_{\text{dMTPA}}^{2}}$$
    (8)

    This guarantees the current vector moves on the \(oA\) trajectory.

  • For a rotor speed beyond the rated speed, the current vector is adjusted as (9) and (10).

    $${i}_{\text{dFW}}={I}_{\text{dMAX}}=\frac{1}{\left({L}_{\text{q}}^{2}-{L}_{\text{d}}^{2}\right)}\left[ \left({L}_{\text{d}}{\lambda }_{\text{PM}}\right)-{(L}_{\text{q}}\cdot \nu )\right]$$
    (9)
    $${i}_{\text{qFW}}={I}_{\text{qMAX}}=\sqrt{{I}_{\text{sMax}}^{2}-{i}_{\text{dFW}}^{2}}$$
    (10)
    $$\mathrm{where}, \nu =\sqrt{{\lambda }_{\text{PM}}^{2}+\left({L}_{\text{q}}^{2}-{L}_{\text{d}}^{2}\right)\left({I}_{\text{sMax}}^{2}-\frac{{V}_{\text{sMax}}^{2}}{{\omega }_{\text{e}}^{2}{L}_{\text{q}}^{2}}\right)}$$
  • Finally, the MTPV operation is activated to allow infinite rotor speed operation, while the voltage constraint (5) is dominant. The current vector is calculated by (11) and (12), i.e. it is forced to follow the \(BC\) trajectory.

    $${i}_{dMTPV}=\frac{1}{{L}_{\text{d}}}\left(\frac{1}{4\left({L}_{\text{q}}-{L}_{\text{d}}\right)} \left({L}_{\text{q}}{\lambda }_{\text{PM}}- \zeta \right)\right)$$
    (11)
    $${i}_{qMTPV}=\frac{1}{{L}_{\text{q}}}\left(\sqrt{{\left(\frac{{V}_{\text{sMax}}}{{\omega }_{\text{e}}}\right)}^{2}-{\lambda }_{\text{PM}}^{2}}\right)$$
    (12)

    where \(\zeta =\sqrt{{{(L}_{\text{q}}{\lambda }_{\text{PM}})}^{2}+8{({L}_{\text{q}}-{L}_{\text{d}})}^{2}{\left(\frac{{V}_{\text{sMax}}}{{\omega }_{\text{e}}}\right)}^{2}}\)

3 Proposed FW control

The proposed algorithm is shown in Fig. 4. A three-level (3L) T-type MLI is used instead of the three-phase VSI. Due to the complex design of classical PI controller with 3L SVPWM, an FCS-MPC is utilized to implement the current control loop for fast tracking of optimal current trajectories. The voltage regulation feedback control is used to adjust the d-axis current component according to the operating region.

Fig. 4
figure 4

Proposed FW control and the T-type MLI

3.1 Model predictive control

MPC is applied as a current control scheme to avoid the limitations of the conventional linear controllers. For example, classical PI controllers have some limitations such as the windup, overshoot, and complex parameters’ tunning and pulse generation. On the other hand, the FCS-MPC strategy utilizes the discrete nature of DSP and microcontrollers to apply an optimal switching state in a very intuitive concept. Hence, measurements of the current period are used to calculate the optimal control vector that provide fast tracking of the d-q axis currents and speed demands.

The implementation process can be summarised as shown in Fig. 5. First, the continuous-time model of the IPM machine, in (1) and (2), is discretized using the first-order Euler discretization. Second, the measured d-q axis currents at the current sample \((k)\) are used to predict the d-q axis currents for the next sample \((k+1)\), as given in (13) [30,31,32]. Third, the optimal voltage vector is selected based on the minimization of the cost function as given in (14). Finally, the switching signals based on the optimal voltage vector are applied to the 3L MLI. It should be noticed that the control set of the proposed FCS-MPC scheme is divided into two groups based on the voltages effect on the neutral point voltage. Therefore, the neutral point voltage is explicitly regulated without including a balancing function in the objective function, and the weighting factor is removed. Furthermore, the number of enumerated vectors is reduced. Thus, the proposed MPC technique is simplified, and it can be easily integrated with the proposed FW strategy.

Fig. 5
figure 5

MPC implementation on the T-type 3L MLI

$$\left.\begin{array}{l}{i}_{\text{sd}}\left(k+1\right)=\left(1-\frac{{R}_{\text{s}}}{{L}_{\text{d}}}{T}_{\text{s}}\right){i}_{\text{sd}}\left(k\right)\\+\left(\frac{{L}_{\text{q}}}{{L}_{\text{d}}}\right){T}_{\text{s}}{i}_{\text{sq}}\left(k\right){\omega }_{\text{e}}\left(k\right)\\+\left(\frac{{T}_{\text{s}}}{{L}_{\text{d}}}\right){u}_{\text{sd}}\left(k\right) \\ {i}_{\text{sq}}\left(k+1\right)=\left(1-\frac{{R}_{\text{s}}}{{L}_{\text{q}}}{T}_{\text{s}}\right){i}_{\text{sq}}\left(k\right)\\-\left(\frac{{L}_{\text{d}}}{{L}_{\text{q}}}{i}_{\text{sd}}\left(k\right)+\frac{{\lambda }_{\text{PM}}}{{L}_{\text{q}}}\right){T}_{\text{s}}{\omega }_{\text{e}}\left(k\right)\\+\left(\frac{{T}_{\text{s}}}{{L}_{\text{q}}}\right){u}_{\text{sq}}\left(k\right)\end{array}\right\}$$
(13)
$$g={({i}_{\text{sd}}^{*}-{i}_{\text{sd}}\left(k+1\right))}^{2}+{({i}_{\text{sq}}^{*}-{i}_{\text{sq}}\left(k+1\right))}^{2}$$
(14)

3.2 Voltage regulation feedback control

The FW control is implemented using the voltage regulation (VR) feedback control [3, 33,34,35].

The voltage regulation feedback control adjusts the d-axis current based on the shortage between the reference voltage vector (\({v}_{\text{sd}}^{*}\) and \({v}_{\text{sq}}^{*}\)) and the maximum inverter voltage \({V}_{\text{sMax}}\). As shown in Fig. 6, the amplitude of the voltage reference voltage is calculated and filtered out using a LPF, and then, it is compared with \({V}_{\text{sMax}}\). The LPF plays two rules; first, it guarantees more stable operation by rejecting the steady-state noise in the voltage feedback loop [36, 37]. In addition, smooth transient is presented by limiting the controller response to the low-frequency component [38].

Fig. 6
figure 6

Voltage regulation feedback control—d-axis current

When the reference voltage vector is higher than \({V}_{\text{sMax}}\), the voltage regulation feedback loop is activated, and the PI controller modifies the d-axis current to follow the FW path (\(AB\)) in Fig. 3. The demagnetizing component is calculated as,

$${i}_{\text{dFW}}={i}_{\text{dMTPA}}+{\Delta i}_{\text{dFW}}$$
(15)

The minimum and maximum limits of \({i}_{\text{dFW}}\) are set to zero and \({I}_{\text{dMAX}}\), where \({I}_{\text{dMAX}}\) is calculated by (16).

$${I}_{\text{dMAX}}={-{I}_{\text{sMax}}-i}_{d(A)}$$
(16)

where \({i}_{d(A)}\) is the d-axis component of point (\(A\)) in Fig. 3 [34].

For the infinite speed IPM machine drives, the MTPV is activated to adjust the d-axis current reference to follow the (\(BC\)) trajectory. This is accomplished by limiting \({i}_{\text{dFW}}\) according to the value of \({i}_{dMTPV}\).

The q-axis component reference is calculated based on the speed controller demand. As shown in Fig. 7, the MTPA is employed to calculate \({i}_{qMTPA}\) based on (8). Two limits are applied to \({i}_{qMTPA}\), the first guarantees that the current reference fulfils the drive’s current limit by limiting \({i}_{qMTPA}\) with \({i}_{\text{qFW}}\), as calculated by (9). The second limit is activated only when the machine operates on the MTPV trajectory. Hence, \({i}_{qMTPV}\), as calculated by (12), is applied to limit \({i}_{qMTPA}\) to the final value of \({i}_{\text{sq}}^{*}\).

Fig. 7
figure 7

Voltage regulation feedback control—q-axis current

It can be seen from Fig. 6 that the VR feedback control operates without the need for machine parameters, and hence, the effect of temperature and/or parameters uncertainty is avoided during the FW region.

4 Simulation results

A MATLAB Simulink model was built to study the performance of the proposed drive as shown in Fig. 4. The IPM, MLI, and controllers’ parameters are given in Table 3. In order to study a realistic system, Prius IPM motor parameters are used in the simulation of this paper [39, 40].

Table 3 Simulation parameters [39, 40]

4.1 MTPA operation

To study the performance of the MTPA operation, the simulation model has been designed to follow the \(OA\) trajectory, i.e. the FW algorithm is not activated. Hence, in this region, the performance of the MLI with two different control techniques, namely the MTPA along with the MPC (MTPA-MPC) and the MTPA along with the SVPWM (MTPA-SV). A reference speed of \({\omega }_{r}^{*}=1.0 \mathrm{krpm}\) is applied to the speed controller, while the load torque is changed from 0 to 250 Nm. This ensures the drive operation below the FW region; meanwhile, the reference current vector moves on the \(OA\) trajectory.

Comparing the two control techniques, as shown in Figs. 8 and 9, shows similar dynamic performance, i.e. the two scenarios fulfil the speed control commands, and the line and d-q axis currents track the reference current commands. However, the torque ripples in the MTPA-MPC technique are lower than that in the MTPA-SV technique due to the decreased ripples of the d-q current components, while the optimal current vector is applied. Also, the sampling interval of the proposed MPC is small enough to accurately track the optimal current vector. This can be confirmed by comparing the FFT analysis of the line current as given in Fig. 10. This FFT analysis has been done at steady-state operation with a load torque of 100 Nm and a rotor speed of \(1.5 \mathrm{krpm}\). The FFT analysis shows reduced harmonics content of the MTPA-MPC technique with a THD of 1.4%. The MTPA-SV algorithm exhibits a higher THD of 3.9%.

Fig. 8
figure 8

Simulation performance of MTPA-MPC

Fig. 9
figure 9

Simulation performance of MTPA-SVPWM

Fig. 10
figure 10

Line current FFT and THD analysis

4.2 FW and MTPV operation

When the system operates in the FW and MTPV regions, the performance of two algorithms can be compared, namely the MPC along with the VR control (MPC-VR) and the MPC along with the optimal current profile (OCP) control (MPC-OCP).

The simulation model is prepared to force the drive operation in both the FW and the MTPV regions by following the \(ABC\) trajectory. Hence, the reference speed is set to a high value of 6.0 krpm, to ensure the drive operation in the FW region. The load torque is decreased in steps from 250 Nm at starting to zero at 0.25 s, and hence, the current vector is forced to move on the \(ABC\) trajectory. Beyond the instant 0.25 s, the load torque is kept at zero to allow maximum acceleration of the motor, and high-speed operation is obtained.

It can be seen, from Figs. 11 and 12, that the proposed MPC-VR algorithm successfully controls the MLI in both the FW and the MTPV regions. The d-q axis current components track the d-q reference components in the FW and MTPV regions and hence the torque-speed curve has been extended compared to the drive operation while the VR algorithm is not applied, see Fig. 13. Also, the MPC-VR provides the same performance as the conventional OCP control (MPC-OCP), and however, cascaded PI speed and current controllers are avoided. Besides, system parameters are necessary to implement the OCP control, and hence, the system performance in practical application is affected by temperature and system parameters uncertainty.

Fig. 11
figure 11

Simulation performance of MPC-VR

Fig. 12
figure 12

Simulation performance of MPC-OCP

Fig. 13
figure 13

Performance comparison

5 Real-time simulation

The complexity of MLI and IPM machine drives imposes fast, safe, and cost-effective design and testing solutions. Recently, real-time (RT) simulations deliver all these features. In addition, system operation with the designated DSPs is confirmed where the RT simulation is performed at the actual operating frequency [41]. The dSPACE DS1103 system is used to test the operation of the T-type MLI and the IPM machine drive. First, the SIMULINK model is prepared for RT simulation [42]. The dSPACE ControlDesk interface, which is shown in Appendix, is employed to process the model, run the RT simulation, and store the results. Second, the SIMULINK model is uploaded to the dSPACE ControlDesk, where the C code for the model is generated. Third, the C code is uploaded to the dSPACE DS1103 board where it is being tested for RT operation. The dSPACE DS1103, the host PC, and the dSPACE CP1103 connection panel are shown in Fig. 14. The host PC has the MATLAB/SIMULINK and the dSPACE ControlDesk installed. The real-time simulation results are shown in Fig. 15 for a fixed load torque of 50 Nm at 1500 rpm, 2000 rpm, 2500 rpm, and 3000 rpm rotor speeds. For the 1500 rpm case, the drive operates at rated speed and the d-q axis current components are calculated accordingly by the MTPA algorithm, while the waveforms show sinusoidal line currents. The 3L operation of line voltage is verified by the phase and line voltage waveforms. The torque and speed fulfil the load and controller demands. The same performance is obtained for the 2000 rpm, 2500 rpm, and 3000 rpm cases, while the d-q axis current components are adapted based on the FW and the MTPV algorithms.

Fig. 14
figure 14

Set-up for the real-time simulation

Fig. 15
figure 15

Real-time simulation results at a fixed load torque of 50 Nm

6 Conclusion

This paper investigates the operation of the T-type MLI-fed IPM machine drive. The VR feedback FW algorithm and the MPC are employed to control the drive in the FW region. The drive operation is compared with the conventional OCP FW control, while the T-type MLI is driven by the SVPWM. The employed MPC scheme utilizes the redundant voltage vectors to balance the dc-link voltage; hence, no weighting factor is required, and reduced enumeration process is implemented. The integration of the proposed FW control with a low-complexity FCS-MPC algorithm has provided several potentials of fast tracking, simple structure, and reduced harmonics thanks to using low sample interval. The proposed drive system has proven effective operation in the MTPA, FW, and MTPV regions. In the FW and MTPV regions, the VR provides the same performance as the OCP control without dependency on system parameters. The system performance has been tested using the MATLAB/SIMULINK and hardware implementation is verified by the dSPACE DS 1103 platform.