Sage Journals: Discover world-class research

Abstract

Traditional finite control set model predictive control (FCS-MPC) selects an optimal voltage vector in one control cycle. However, considering multiple control cycles, it cannot be proved that the voltage vector is optimal, and the uncertainty of PMSM parameters seriously affects the prediction accuracy. In order to balance the relationship between switching frequency, steady-state performance, and robustness, a multi-step robust FCS-MPC is proposed. Firstly, an incremental prediction model is established to eliminate the flux linkage of permanent magnets. The inductance parameters in the incremental model are identified online based on the observer and inductance extraction algorithm. Then, based on the principle of no-beat, the candidate voltage vector in two control periods is simplified to get the candidate voltage vector, and the cost function is used to determine the optimal voltage vector again. Finally, the proposed FCS-MPC method is compared with the traditional FCS-MPC method. The experimental results show the effectiveness of FCS-MPC strategy.

Keywords

Permanent magnet synchronous motor multi-step model predictive control parameter mismatch robust control

Introduction

With the rapid development of industry, society is facing serious problems such as energy shortage and environmental pollution. Therefore, it is very necessary to develop new clean energy. Many countries and regions focus on wind power generation, new energy vehicles, and other aspects.^1–6 As the core component of the above system, the motor has been widely studied, among which, the permanent magnet synchronous motor stands out in a crowd of motors because of its high-power density, high efficiency, lightweight, and other advantages.^7–11

The control algorithm directly affects the performance of the PMSMs system. The traditional schemes such as magnetic field-oriented control (FOC) and direct torque control (DTC) have the disadvantages of insufficient dynamic performance and poor steady-state performance respectively.^12–17 In recent years, model predictive control (MPC) has been regarded as a powerful alternative to FOC because of its simple principle, easy of dealing with multiple constraints, and fast dynamic response.^18–21 According to different control variables, MPC can be divided into predictive current control,^22–24 predictive torque control,^25,26 and predictive speed control,^27,28 among which predictive current control is widely used because there is no adjustment of the weighting factor. According to whether the modulation scheme is adopted or not, MPC can be divided into finite control set model predictive control (FCS-MPC)^29,30 and continuous control set model predictive control (CCS-MPC),³¹ wherein FCS-MPC can calculate the optimal voltage vector by minimizing the cost function and directly act on the inverter, and it is easy to add constraints, which is more concise.³²

Conventional FCS-MPC outputs only one voltage vector in one control cycle, which inevitably leads to a large error between the expected voltage and the output voltage and deteriorates the steady-state performance. To solve this problem, a multi-vector scheme is proposed.³³ The basic principle of the multi-vector scheme is to improve the steady-state performance of the system by outputting multiple voltage vectors in one control period. Duty cycle control is used in Sun et al.,³⁴ zero vector and non-zero voltage vector are output in a control period so that the amplitude of the output voltage is adjusted. In Gonçalves et al.,³⁵ the output voltage amplitude is adjusted by constructing a virtual voltage vector. In Sun et al.,³⁶ duty cycle control and virtual vector are integrated to adjust the amplitude and phase of the output voltage simultaneously. In addition, two-vector MPC and three-vector MPC exist,^37–39 and the reference modulation scheme greatly increases the flexibility of the output voltage. However, the above scheme undoubtedly increases the average switching frequency of the system because of the output of multiple voltage vectors in one control cycle.

To balance the relationship between the steady-state performance of FCS-MPC and the switching frequency, a multi-step FCS-MPC is introduced. The multi-step FCS-MPC can reduce the switching frequency and improve the steady-state performance by considering the optimization of multiple control cycles and adding switching limits to the cost function.^40–42 However, the set of alternative voltage vectors that need to be computed iteratively for multi-step FCS-MPC increases exponentially with the increase in step size. Aiming at the heavy computation burden, many scholars reduce the computation amount by using a spherical decoding algorithm or gradient descent scheme.^43,44 A simple two-step scheme is proposed in Zhang et al.⁴⁵ The series structure is used to first reduce the alternative voltage vector set and then the cost function is used to calculate the optimal voltage vector, which reduces the switching frequency and improves the steady-state performance. However, another problem with FCS-MPC is that it has a strong parameter dependence. The prediction accuracy will decrease with parameter mismatch.

At present, many scholars have studied the robustness of model prediction, which can be divided into observer-based estimation compensation, online parameter identification, and model-free scheme.^30,46 The principle of an observer-based estimation compensation scheme is to estimate and compensate the disturbance caused by parameter mismatch as a whole to the reference voltage. In Sun et al.,⁴⁷ a sliding mode observer is used to estimate the lumped disturbance compensation to the reference voltage, and a robust dead-beat predictive control is realized. Li et al.⁴⁸ applies the observer to FCS-MPC, and the estimated disturbance compensation to the reference voltage is used to determine the optimal voltage vector. At the same time, the virtual voltage vector is added to improve the accuracy of the voltage output, effectively reducing the parameter sensitivity of FCS-MPC. Another efficient scheme is to estimate time-varying parameters online and apply them to the prediction model. Traditional identification schemes include recursive least square (RLS) method, extended Kalman filter (EKF), and model reference adaptive (MRA) method.^49–53 RLS requires a lot of computing and storage space. The problem of adaptive rule design is complicated in MRA. EKF has a lot of matrix operations and it is also very complex to identify multiple parameters. The advantage of a model-free scheme is that it does not use motor parameters to avoid the impact of parameter mismatch, which has attracted the attention and research of many scholars. Model-free schemes include the use of multiple sampling current differences to predict future currents,⁵⁴ or the design of a hyperlocal model using a first-order hyperlocal model instead of the PMSM model.^55,56 In addition, the model-free scheme also includes a data-driven approach to design the controller directly through input and output,⁵⁷ but there are problems such as a large amount of required data, difficult data processing, and a large computational burden.

Therefore, a robust two-step FCS-MPC is proposed. The incremental prediction model is constructed to eliminate the flux linkage of the permanent magnet and reduce the voltage vector backup set by the principle of no beat and voltage vector non-switching. Inductance parameters in the incremental prediction model are identified online by designing an observer and an inductance extraction algorithm. Finally, the optimal voltage vector is obtained by substituting the reduced voltage vector set into the cost function.

The rest of this article is organized as follows. Section “Conventional FCS-MPC of PMSMs” describes the model of permanent magnet synchronous motors and the traditional FCS-MPC. Section “Simplified robust two-step FCS-MPC” introduces the simplified two-step robust FCS-MPC. Section “Experimental result” shows a comparison of the proposed scheme with conventional FCS-MPC. Conclusions and future outlook are presented in Section “Conclusion” and Section “Future outlook,” respectively.

Conventional FCS-MPC of PMSMs

Mathematical model of the PMSMs

In this paper, surface-mounted PMSMs (SPMSMs) are used, and their inductance meets L_d = L_q = L. Therefore, the voltage equation of SPMSMs in the rotating coordinate system (d−q coordinate system) is expressed as:

u = A i + B \frac{d}{d t} i + C

(1)

where

\begin{matrix} u = [\begin{matrix} u_{d} \\ u_{q} \end{matrix}], i = [\begin{matrix} i_{d} \\ i_{q} \end{matrix}], A = [\begin{matrix} R_{s} & - ω_{e} L \\ ω_{e} L & R_{s} \end{matrix}], \\ B = [\begin{matrix} L & 0 \\ 0 & L \end{matrix}], C = [\begin{matrix} 0 \\ ω_{e} ψ_{f} \end{matrix}] \end{matrix}

u_d(q), i_d(q) represent the voltage component and the current component of the d−q axis respectively, and ωe is the electric angular velocity. R_s and ψ_f and represent stator resistance and flux linkage.

The electromagnetic torque equation of the SPMSMs is given by

T_{e} = \frac{3}{2} P_{n} i_{q} ψ_{f}

(2)

where Pn represents the number of pole pairs of SPMSMs.

Conventional FCS-MPC strategy

According to Euler’s discrete equation, (1) can be expressed as:

i (k + 1) = A_{1} i (k) + B_{1} u (k) + C_{1}

(3)

where

A_{1} = [\begin{matrix} 1 - \frac{T_{c} R_{s}}{L} & T_{c} ω_{e} \\ - T_{c} ω_{e} & 1 - \frac{T_{c} R_{s}}{L} \end{matrix}], B_{1} = [\begin{matrix} \frac{T_{c}}{L} & 0 \\ 0 & \frac{T_{c}}{L} \end{matrix}], C_{1} = [\begin{matrix} 0 \\ - \frac{T_{c} ω_{e} ψ_{f}}{L} \end{matrix}]

T_c is the control period, k and k + 1 represent the current moment and the next moment, respectively.

The eight basic voltage vectors of the two-level voltage source inverter (2L-VSI) shown in Figure 1 are substituted into (3), and the result of the predicted current is compared with the reference value. By minimizing the current tracking error, the following cost function is established:

\begin{matrix} J = {(i^{ref} - i (k + 1))}^{2} \\ s . t . u \in {u_{0}, u_{1}, . . ., u_{7}} \end{matrix}

(4)

where the superscript “ref” represents the reference value. The reference value of i_d is 0, and the reference value of i_q is output by the PI controller of the speed loop. The block diagram of conventional FCS-MPC is shown in Figure 2.

Figure 1.

The basic voltage vectors of 2L-VSI.

Figure 2.

Block diagram of conventional FCS-MPC.

Based on the above analysis, the traditional FCS-MPC only outputs one voltage vector in one control cycle, resulting in poor steady-state performance. An effective solution to this problem is multi-vector MPC. The application performance of a multi-vector scheme in FCS-MPC is given in Li et al.,³⁰ and its performance is shown in Figure 3. It can be observed that the current THD and torque ripple of the system decreases with the increase of the vector output, but the calculation time of the scheme is undoubtedly further extended. In addition, as the switching of voltage vectors becomes faster, the average switching frequency of the system undoubtedly increases.

Figure 3.

Performance comparison of multi-vector schemes in Li et al.³⁰

Simplified robust two-step FCS-MPC

The control block diagram of the simplified two-step robust FCS-MPC strategy proposed in this paper is shown in Figure 4, which mainly includes step1 and step2: reference voltage calculation and candidate voltage vector set determination based on the deadbeat principle, incremental prediction model, cost function, and inductance extraction algorithm.

Figure 4.

Block diagram of the simplified two-step robust FCS-MPC.

Incremental prediction model

According to (3), the accuracy of the prediction model depends on the accuracy of the motor parameters. To reduce the sensitivity of parameters, an incremental prediction model was established to eliminate the influence of permanent magnet flux on the prediction formula. The specific steps are as follows:

First, a backward Euler expansion of (1) is performed to obtain:

i (k) = A_{1} i (k - 1) + B_{1} u (k - 1) + C_{1}

(5)

Then, subtract (5) from (3) to obtain the incremental prediction current model:

Δ i (k + 1) = A_{1} Δ i (k) + B_{1} Δ u (k)

(6)

where

Δ i (k) = [\begin{matrix} i_{d} (k) - i_{d} (k - 1) \\ i_{q} (k) - i_{q} (k - 1) \end{matrix}]; Δ u (k) = [\begin{matrix} u_{d} (k) - u_{d} (k - 1) \\ u_{q} (k) - u_{q} (k - 1) \end{matrix}]

Due to the calculation delay, the current prediction model obtained by one-step delay compensation for model (6) is as follows:

Δ i (k + 2) = A_{1} Δ i (k + 1) + B_{1} Δ u (k + 1)

(7)

If traditional iterative calculation is used, eight calculations are required to determine the optimal voltage vector. To reduce the computational burden, based on the deadbeat principle, the predicted current i_d(k + 2) and i_q(k + 2) should reach their reference values by the end of the next control cycle. Therefore, i (k + 2) is replaced by the current reference value i ^ref, so the predicted reference voltage under the incremental model is expressed as:

\begin{matrix} [\begin{matrix} {u_{d}}^{ref} \\ {u_{q}}^{ref} \end{matrix}] = {B_{1}}^{- 1} [\begin{matrix} {i_{d}}^{ref} - i_{d} (k + 1) \\ {i_{q}}^{ref} - i_{q} (k + 1) \end{matrix}] \\ - {B_{1}}^{- 1} A_{1} [\begin{matrix} i_{d} (k + 1) - i_{d} (k) \\ i_{q} (k + 1) - i_{q} (k) \end{matrix}] + [\begin{matrix} u_{d} (k) \\ u_{q} (k) \end{matrix}] \end{matrix}

(8)

Simplified two-step MPC

Unlike traditional FCS-MPC, which only considers the current tracking within one control period and determines an optimal voltage vector after obtaining the reference voltage, the prediction range is extended to two steps. When the optimal voltage vectors of two adjacent control cycles are the same, the switching state of the inverter remains unchanged, that is, the switching change is suppressed. Therefore, to balance switching frequency and steady-state performance, the selection of the optimal voltage vector is extended to two control cycles.

Therefore, the proposed voltage vector optimization scheme is as follows: First, after the reference voltage is obtained, it is converted to the reference voltage in the rest coordinate system according to the change of coordinates. Then, according to the reference voltage, the backup set of the optimal voltage vector is determined in the voltage vector space as shown in Figure 5. The relationship between sectors and alternative voltage vector sets is shown in Table 1. Each sector contains one zero vector and two non-zero vectors, and through the principle of no beat, the calculation is reduced from eight times to one time, which effectively reduces the calculation burden. It is worth noting that different from the traditional reference voltage calculation scheme based on the principle of no beat, the reference voltage scheme in this paper can not directly determine the optimal voltage vector, and it is only used for the preselection of the optimal voltage vector within two control cycles.

Figure 5.

Voltage positioning method based on deadbeat principle.

Table 1.

Relation between sectors and alternative voltage vector sets.

Sector	Alternative voltage vector set
1	u ₀(u₇), u₄, u₆
2	u ₀(u₇), u₂, u₆
3	u ₀(u₇), u₃, u₂
4	u ₀(u₇), u₃, u₁
5	u ₀(u₇), u₁, u₅
6	u ₀(u₇), u₄, u₅

After the alternative voltage vector set is obtained, the second step is to select the optimal voltage vector from the three candidate voltage vectors. The optimal voltage vector backup set is the set of optimal and suboptimal voltage vectors at the k + 2 moment. The principle of the series structure proposed is to evaluate the alternative voltage vector set at k + 2 moment by current tracing at k + 3 moment. Therefore, the scheme can be understood as the selection of the optimal set of voltage vectors within two steps.

The method of selecting the optimal voltage vector set at k + 2 moment is introduced in (8). The predicted current increment model at k + 3 moment is expressed as follows:

Δ i (k + 3) = A_{1} Δ i (k + 2) + B_{1} Δ u (k + 2)

(9)

Then, taking sector 1 as an example, the corresponding cost function based on the current tracking error is designed. Select three candidate voltage vectors from the k + 2 moment and plug them into (9) to obtain the optimal voltage vector output to the inverter.

\begin{matrix} J = {(i^{ref} - i (k + 3))}^{2} \\ s . t . u \in {u_{0} (u_{7}), u_{4}, u_{6}} \end{matrix}

(10)

2L-VSl has two voltage vectors of zero amplitude, to reduce the average switching frequency of the system. The rules for applying the two zero vectors are as follows: If the optimal output at the previous time is u₄ (whose switching signal is Sa = 1, Sb = 0, Sc = 0), then u₀ (whose switching signal is Sa = 0, Sb = 0, Sc = 0) is used instead of u₇ (whose switching signal is Sa = 1, Sb = 1, Sc = 1). If the optimal output voltage vector at the previous time is u₆ (whose switching signal is Sa = 1, Sb = 1, Sc = 1), then u₇ is used (whose switching signal is Sa = 1, Sb = 1, Sc = 1).

Inductance extraction algorithm

The incremental prediction model shown in (7) eliminates the influence of permanent magnet flux linkage, but inductance and resistance parameters still exist, and the influence caused by the mismatch of resistance parameters is small and can be ignored.⁵⁸ Therefore, an inductance extraction algorithm is designed to enhance the robustness of the SPMSMs drive system.

Since the d-axis inductance and the q-axis inductance are equal, taking the d-axis as an example, the voltage equation considering parameter uncertainty is expressed as:

{\begin{matrix} u_{d} = R_{s} i_{d} + L \frac{d i_{d}}{dt} - ω_{e} L i_{q} + ε_{d} \\ ε_{d} = Δ R_{s} i_{d} + Δ L \frac{d i_{d}}{dt} - ω_{e} Δ L i_{q} \\ \frac{d ε_{d}}{dt} = δ_{d} \end{matrix}

(11)

The perturbation term is estimated by designing an observer:

{\begin{matrix} u_{d} = R_{s} {\hat{i}}_{d} + L \frac{d {\hat{i}}_{d}}{dt} - ω_{e} L i_{q} + {\hat{ε}}_{d} + f_{d} \\ \frac{d {\hat{ε}}_{d}}{dt} = c_{d} f_{d} \end{matrix}

(12)

where “^” represents the estimated value. f_d is the control function, c_d is a positive sliding mode parameter.

Subtract (11) from (12) to obtain the estimated error term:

{\begin{matrix} \frac{d e_{1}}{dt} = - \frac{1}{L} (R_{s} e_{1} + e_{2} + f_{d}) \\ \frac{d e_{2}}{dt} = c_{d} f_{d} - δ_{d} \end{matrix}

(13)

where $e_{1} = {\hat{i}}_{d} - i_{d}, e_{2} = {\hat{ε}}_{d} - ε_{d}$ .

The sliding mode function adopts variable gain approximation law to reduce tremor, expressed as:

\frac{ds}{dt} = - k_{1} s - \frac{k_{2}}{(1 - δ) e^{- β | s |} + δ} sign (s)

(14)

The stability analysis of the designed sliding mode observer is as follows:

\frac{dV}{dt} = \frac{ds}{dt} \cdot s = - k_{1} s^{2} - \frac{k_{2}}{(1 - δ) e^{- β | s |} + δ} s sign (s) \leq 0

(15)

When the parameters k₁ > 0, k₂ > 0, β > 1, and 0 < δ < 1 are set, the reaching law satisfies the stability condition, which means that the system is stable.

Taking e₂ as the disturbance of the control function, design the following sliding mode control function as follows:

f_{d} = L [k_{1} e_{1} + \frac{k_{2}}{(1 - δ) e^{- β | e_{1} |} + δ} sign (e_{1})] - R_{s} e_{1}

(16)

Substitute (16) into (12) and e₁ is considered equal to zero at steady state, so it can be simplified to:

\frac{d {\hat{ε}}_{d}}{dt} = c_{d} f_{d} = c_{d} \hat{L} k_{2} sign (e_{1})

(17)

where $\hat{L}$ is the estimated value of inductance.

According to (16), the estimated inductance can be derived as:

\hat{L} = | \frac{d {\hat{ε}}_{d}}{dt} \frac{1}{c_{d} k_{2}} |

(18)

The inductance extraction algorithm is shown in Figure 6, where k_p and k_i are the proportional and integral gains of the PI controller. The transfer function is expressed in the form of a typical type II system:

G (s) = \frac{c_{d} k_{2} k_{i} (\frac{k_{p}}{k_{i}} + 1)}{s^{2} (s + 1)}

(19)

Figure 6.

The inductance extraction algorithm.

Experimental result

Figure 7 shows the construction of the experimental device, and the relevant SPMSM parameters are shown in Table 2. To verify the advantages of the scheme in steady-state performance, switching frequency, and robustness, the proposed scheme is compared with conventional FCS-MPC. Rtunit’s real-time controller RTU-BOX 201 was adopted in the experiment, which integrated ADC, PWM, position signal input, and high-quality communication interface modules, and the experimental sampling frequency was set at 10 kHz. The controller block diagram based on the experiment of the proposed scheme is shown in Figure 8.

Figure 7.

The experiment platform.

Table 2.

SPMSM system parameters.

Parameters	Values
Inductance	6.6 mH
Resistance	1.6 Ω
Number of pole pairs	4
Rated speed	2500 rpm
Rated power	3.8 kW
Rated torque	15 Nm
Inertia	0.00277 kg m²

Figure 8.

Controller block diagram based on RTU BOX.

Steady-state performance and switching frequency comparison under normal parameters

First, to demonstrate that a simplified two-step FCS-MPC can effectively trade off the relationship between switching frequency and steady-state performance. The steady-state performance and switching frequency of conventional FCS-MPC and simplified two-step FCS-MPC are compared under normal parameters. First, the control frequency of both schemes is set to 20 kHz. Set the load torque to 8 Nm and the speed to 2000 rpm. Figure 9 shows the speed, phase current, and corresponding harmonic analysis of the two schemes. It can be seen from Figure 9 that both schemes can track the reference speed well, with the THD of the traditional scheme being 5.51% and that of the proposed scheme being 7.26%. To illustrate the universality of the scheme, the THD of the two schemes was compared under the same load but at different speeds, such as 500, 1000, and 1500 rpm, as shown in Figure 10. Figure 10 illustrates that the steady-state performance of conventional FCS-MPC is slightly better than that of the proposed two-step FCS-MPC. However, this is shown at different switching frequencies. At the same time, the switching frequency of the two schemes is shown in Figure 11, and it can be observed that the switching frequency of the proposed scheme is always lower than that of the traditional FCS-MPC.

Figure 9.

Comparison of speed, A-phase current, and THD: (a) conventional FCS-MPC and (b) the proposed FCS-MPC.

Figure 10.

Comparison of THD with different motor speeds.

Figure 11.

Comparison of switching frequency with different motor speeds.

To show the superiority of the proposed scheme, the THD of the traditional FCS-MPC and the proposed scheme are compared by maintaining the same switching frequency at the same speed of 2000 rpm and load of 8 Nm. To prove its universality, it was verified at different rotational speeds, as shown in Figures 12 and 13. According to the data in Figures 12 and 13, it can be seen that the proposed scheme has smaller THD and better steady-state performance under the same switching frequency and experimental conditions. Therefore, the proposed two-step FCS-MPC can effectively balance the relationship between steady-state performance and switching frequency.

Figure 12.

THD of traditional FCS-MPC at 2000 rpm, 8 Nm load, and 2.31 kHz switching frequency.

Figure 13.

THD comparison of two schemes at the same switching frequency and different speeds.

In addition, to further prove the effectiveness of the proposed scheme. The proposed scheme is compared with the two-step FCS-MPC proposed in Zhang et al.,⁵⁹ which applies the same voltage vector in both control cycles. The motor speed is set at 1000 rpm and the load is set at 10 Nm. Figure 14 shows the speed and current THD of the two schemes.

Figure 14.

Speed and THD comparison of the proposed FCS-MPC and FCS-MPC in Zhang et al.⁵⁹

It can be observed that the THD of the proposed scheme is smaller, and the THD of the two is 5.98% and 4.66% respectively, because the vector is not restricted in the two control cycles of the proposed scheme, while the FCS-MPC in Zhang et al.⁵⁹ considers the vector invariant in the two control cycles. Table 3 shows the calculation time and the average switching frequency of the system for the above two schemes and the traditional FCS-MPC. Because the switches in the two control cycles are forcibly restricted, the average switching frequency in Zhang et al.⁵⁹ is smaller, but the calculation time is relatively maximum. The calculation burden is large because 11 enumerations are required in Zhang et al.,⁵⁹ including eight enumerations for the first time for the preselection of three voltage vectors and three enumerations for the second time for the selection of the optimal value. The scheme proposed in this paper requires three enumerations, while the traditional FCS-MPC requires eight enumerations.

Table 3.

Comparison of average switching frequency and calculation time between the proposed FCS-MPC and FCS-MPC in Zhang et al.⁵⁹

Methods	Average switching frequency (kHz)	Calculation time (μs)
Conventional FCS-MPC	2.85	45
FCS-MPC in Zhang et al.⁵⁹	2.54	48
The proposed FCS-MPC	2.64	40

Robust performance comparison

The parameters of the sliding mode observer and inductance extraction algorithm are set continuously as follows: k₁ = 800, k₂ = 800, β = 2, δ = 0.3, k_p = 0.04, k_i = 0.03. To show the suppression of parameter sensitivity by the proposed incremental prediction model and inductance algorithm, the conventional FCS-MPC is compared with the proposed two-step FCS-MPC. Because the inductance parameters of the motor cannot be changed quantitatively in real time, this paper simulates the disturbance under the change of parameters by changing the inductance parameters in the controller. First, because the variable φf is not used in the incremental prediction model, it will not be affected by its parameter mismatch. At 0.7 s, the load torque is set to rise from 5 to 8 Nm, and the speed is set to 1500 rpm. Figures 15 and 16 show the stator current tracking effects of the two schemes under 200% and 50% inductance respectively. The red realization represents the reference value of the current, and the dashed line represents the time point at which the inductance changes. It can be seen from Figures 15 to 16 that traditional FCS-MPC has large tracking errors when inductance parameters are mismatched, while the proposed robust two-step FCS-MPC has good current tracking performance. To further explain the superiority of the current tracking performance of the proposed scheme, the tracking errors of both are shown in Table 4. It can be seen from Table 4 that when the parameters are mismatched, the tracking error of the proposed scheme is small, no matter the d-axis current or the q-axis current. Therefore, the experimental results demonstrate the robustness of the proposed scheme.

Figure 15.

Comparison of current tracking performance at 200%L: (a) conventional FCS-MPC and (b) the proposed FCS-MPC.

Figure 16.

Comparison of current tracking performance at 50%L: (a) conventional FCS-MPC and (b) the proposed FCS-MPC.

Table 4.

Comparison of current tracking error between two schemes.

Methods	Parameter mismatch	Current tracing error
Methods	Parameter mismatch	i_d (A)	i_q (A)
Conventional FCS-MPC	200%L	0.2354	0.6541
The proposed FCS-MPC		0.0332	0.3012
Conventional FCS-MPC	50%L	0.3713	0.5377
The proposed FCS-MPC		0.1235	0.2315

Conclusion

In this paper, a simplified robust two-step FCS-MPC is proposed, which extends the optimization of traditional FCS-MPC from one step to two control cycles. In this study, an incremental prediction model is established to eliminate the influence of permanent magnet flux linkage. On this basis, the reference voltage is calculated using the principle of deadbeat control, and three alternative voltage vectors are determined by the principle of non-switching voltage vectors within two steps. After obtaining the alternative vector set, the optimal voltage vector is calculated by using the cost function of the series structure, and the output is applied, which greatly reduces the computational burden. In addition, inductance parameters still affect the accuracy of the prediction model. A sliding mode observer and related inductance extraction algorithm are designed to optimize the incremental prediction model. By comparing the experimental results, the advantages of the proposed scheme in average switching frequency, steady-state performance, and robustness performance are verified.

Future outlook

Compared with traditional FCS-MPC, the proposed two-step scheme has improved the average switching frequency, steady-state performance, and computational cost, but the performance optimization is not great when the step size is less than three steps. The author will consider the three-step FCS-MPC in the future and will calculate the reference voltage of the three control periods based on this paper by extending the deadbeat principle to three steps and selecting the optimal voltage vector sequence in the voltage plane of the three control periods according to the reference voltage.

Footnotes

Handling Editor: Sharmili Pandian

Author contributions

Conceptualization, C.Z.; formal analysis, C.Z. and C.C; investigation, C.Z. and C.C; writing – original draft preparation, C.Z.; All authors have read and agreed to the published version of the manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Institutional review board statement

Not applicable.

Informed consent statement

Not applicable.

ORCID iDs

Chen Zhang

Chen Cen

Data availability statement

Not applicable.

References

Sun

Dong

Jin

, et al. System-level energy management optimization of power-split hybrid electric vehicle based on nested design. IEEE Trans Ind Electron. Epub ahead of print 15 December 2023. DOI: 10.1109/TIE.2023.3340212.

Zhang

Chen

You

, et al. Energy-aware optimization of connected and automated electric vehicles considering vehicle-traffic nexus. IEEE Trans Ind Electron 2024; 71: 282–293.

Wang

Zhang

Cao

, et al. Modeling and detection scheme for zero-dynamics attack on wind power system. IEEE Trans Smart Grid 2024; 15: 934–943.

Sun

Jin

Xue

, et al. Adaptive ECMS with gear shift control by grey wolf optimization algorithm and neural network for plug-in hybrid electric buses. IEEE Trans Ind Electron 2024; 71: 667–677.

Sun

Xue

Cai

, et al. Adaptive ECMS based on EF optimization by model predictive control for plug-in hybrid electric buses. IEEE Trans Transp Electrif 2023; 9: 2153–2163.

Sun

Yao

, et al. Efficient feedback linearization control for an IPMSM of EVs based on improved firefly algorithm. ISA Trans 2023; 134: 431–441.

Bai

Yin

Zhang

, et al. Multiple-models adaptive disturbance observer-based predictive control for linear permanent-magnet synchronous motor vector drive. IEEE Trans Power Electron 2022; 37: 9596–9611.

Shi

Sun

Yang

, et al. Design optimization of a spoke type axial-flux PM machine for in-wheel drive operation. IEEE Trans Transp Electrif. Epub ahead of print 31 August 2023. DOI: 10.1109/TTE.2023.3310738.

Sun

Yao

. Sequential subspace optimization design of a dual three-phase permanent magnet synchronous hub motor based on NSGA III. IEEE Trans Transp Electrif 2023; 9: 622–630.

10.

Guo

, et al. Fault-tolerant current control of six-phase permanent magnet motor with multifrequency quasi-proportional-resonant control and feedforward compensation for aerospace drives. IEEE Trans Power Electron 2023; 38: 283–293.

11.

Shi

Sun

Lei

, et al. Multiobjective optimization of a five-phase bearingless permanent magnet motor considering winding area. IEEE ASME Trans Mechatron 2022; 27: 2657–2666.

12.

Petkar

Thippiripati

. A novel duty-controlled DTC of a surface PMSM drive with reduced torque and flux ripples. IEEE Trans Ind Electron 2023; 70: 3373–3383.

13.

Jin

Sun

Lei

, et al. Sliding mode direct torque control of SPMSMs based on a hybrid wolf optimization algorithm. IEEE Trans Ind Electron 2022; 69: 4534–4544.

14.

Chikondra

Muduli

Behera

. An improved open-phase fault-tolerant DTC technique for five-phase induction motor drive based on virtual vectors assessment. IEEE Trans Ind Electron 2021; 68: 4598–4609.

15.

Candelo-Zuluaga

Riba

Garcia

. PMSM parameter estimation for sensorless FOC based on differential power factor. IEEE Trans Instrum Meas 2021; 70: 1–12.

16.

Tian

Q-T

Duan

J-D

, et al. Cancellation of torque ripples with FOC strategy under two-phase failures of the five-phase PM motor. IEEE Trans Power Electron 2017; 32: 5459–5472.

17.

Sun

Xiong

Yao

, et al. A hybrid control strategy for multimode switched reluctance motors. IEEE ASME Trans Mechatron 2022; 27: 5605–5614.

18.

Yang

Miao

Sun

. Model predictive current control for IPMSM drives with extended-state-observer-based sliding mode speed controller. IEEE Trans Energy Convers 2023; 38: 1471–1480.

19.

Vazquez

Rodriguez

Rivera

, et al. Model predictive control for power converters and drives: advances and trends. IEEE Trans Ind Electron 2017; 64: 935–947.

20.

Andersson

Thiringer

. Assessment of an improved finite control set model predictive current controller for automotive propulsion applications. IEEE Trans Ind Electron 2020; 67: 91–100.

21.

Sun

Lei

, et al. Robust predictive cascaded speed and current control for PMSM drives considering parameter variations. IEEE Trans Ind Electron. Epub ahead of print 8 December 2023. DOI: 10.1109/TIE.2023.3337498.

22.

Sun

Tian

, et al. Fault-tolerant operation of a six-phase permanent magnet synchronous hub motor based on model predictive current control with virtual voltage vectors. IEEE Trans Energy Convers 2021; 37: 337–346.

23.

Gong

Ding

, et al. Flux observation transfer-based inductance identification technique for precise FCS-MPCC used in surface-mounted PMSMs. IEEE Trans Power Electron 2023; 38: 4241–4245.

24.

Sun

Lin

Guo

, et al. Improved deadbeat predictive current control with extended state observer for dual three-phase PMSMs. IEEE Trans Power Electron 2024; 39: 6769–6782.

25.

Zhang

Zhu

Zhao

, et al. An improved direct torque control for three-level inverter-fed induction motor sensorless drive. IEEE Trans Power Electron 2012; 27: 1502–1513.

26.

Sun

Zhu

, et al. Improved model predictive torque control for PMSM drives based on duty cycle optimization. IEEE Trans Magn 2021; 57: 1–5.

27.

Wang

Luo

Chen

, et al. An improved robust model predictive speed control with inertia identification for PMSM drives in electro-hydrostatic actuator. IEEE Trans Power Electron 2023; 38: 13825–13841.

28.

Wang

Rodríguez

. A robust predictive speed control for SPMSM systems using a sliding mode gradient descent disturbance observer. IEEE Trans Energy Convers 2023; 38: 540–549.

29.

Liu

Hameyer

. Torque ripple minimization for direct torque control of PMSM with modified FCSMPC. IEEE Trans Ind Appl 2016; 52: 4855–4864.

30.

Sun

Lei

, et al. Finite-control-set model predictive control of permanent magnet synchronous motor drive systems—an overview. IEEE CAA J Automatica Sinica 2022; 9: 2087–2105.

31.

Mohammed

SAQ

Alsofyani

Lee

. Improved adaptive CCS-MPCC for distorted model parameters mitigation of IPMSM drives. IEEE Trans Ind Electron 2024; 71: 3483–3493.

32.

Ahmed

Koh

Lee

. A comparison of finite control set and continuous control set model predictive control schemes for speed control of induction motors. IEEE Trans Industr Inform 2018; 14: 1334–1346.

33.

Yao

Sun

. Fault-tolerant control of five-phase permanent magnet synchronous hub motor based on improved model predictive current control. J Electr Eng Technol 2023; 19: 247–257.

34.

Sun

Zhu

, et al. Speed sensorless model predictive current control based on finite position set for PMSHM drives. IEEE Trans Transp Electrif 2021; 7: 2743–2752.

35.

Gonçalves

PFC

Cruz

SMA

Mendes

AMS

. Multistage predictive current control based on virtual vectors for the reduction of current harmonics in six-phase PMSMs. IEEE Trans Energy Convers 2021; 36: 1368–1377.

36.

Sun

Yao

, et al. Improved finite-control-set model predictive control with virtual vectors for PMSHM drives. IEEE Trans Energy Convers 2021; 37: 1885–1894.

37.

Farah

Lei

Zhu

, et al. Two-vector dimensionless model predictive control of PMSM drives based on fuzzy decision making. CES Trans Electr Mach Syst 2022; 6: 393–403.

38.

Cheng

Jianhui

Jing

. Dual-vector predictive current control of open-end winding PMSM with zero-sequence current hysteresis control. IEEE J Emerg Sel Top Power Electron 2022; 10: 184–195.

39.

Zhang

Huang

. Generalized multiple-vector-based model predictive control for PMSM drives. IEEE Trans Ind Electron 2018; 65: 9356–9366.

40.

Jiang

Yang

Fan

, et al. An improved implicit model predictive current control with continuous control set for PMSM drives. IEEE Trans Transp Electrif 2022; 8: 2444–2455.

41.

Geyer

Quevedo

. Multistep direct model predictive control for power electronics—part 2: analysis. In: Proceedings of 2013 IEEE energy conversion congress and exposition, Denver, CO, USA, 15–19 September 2013, pp.1162–1169. New York: IEEE.

42.

Geyer

Quevedo

. Multistep direct model predictive control for power electronics—part 1: algorithm. In: Proceedings of 2013 IEEE energy conversion congress and exposition, Denver, CO, USA, 15–19 September 2013, pp.1154–1161. New York: IEEE.

43.

Grimm

Kolahian

Zhang

, et al. A sphere decoding algorithm for multistep sequential model-predictive control. IEEE Trans Ind Appl 2021; 57: 2931–2940.

44.

Wang

Xie

Wang

, et al. Fast response model predictive torque and flux control with low calculation effort for PMSMs. IEEE Trans Industr Inform 2019; 15: 5531–5540.

45.

Zhang

Bai

Cheng

. Improved model predictive current control with series structure for PMSM drives. IEEE Trans Ind Electron 2022; 69: 12437–12446.

46.

Rodriguez

Garcia

Mora

, et al. Latest advances of model predictive control in electrical drives—part I: basic concepts and advanced strategies. IEEE Trans Power Electron 2022; 37: 3927–3942.

47.

Sun

Zhang

Cai

, et al. Compensated deadbeat predictive current control considering disturbance and VSI nonlinearity for in-wheel PMSMs. IEEE ASME Trans Mechatron 2022; 27: 3536–3547.

48.

Sun

Yao

, et al. Improved finite control set model predictive current control for permanent magnet synchronous motor with sliding mode observer. IEEE Trans Transp Electrif 2024; 10: 699–710.

49.

Sun

Zhang

Tian

, et al. Speed sensorless control for IPMSMs using a modified MRAS with gray wolf optimization algorithm. IEEE Trans Transp Electrif 2022; 8: 1326–1337.

50.

Brosch

Wallscheid

Böcker

. Long-term memory recursive least squares online identification of highly utilized permanent magnet synchronous motors for finite-control-set model predictive control. IEEE Trans Power Electron 2023; 38: 1451–1467.

51.

Kashif

Singh

. Reduced-sensor-based multistage model reference adaptive control of PV-fed PMSM drive for water pump. IEEE Trans Ind Electron 2023; 70: 3782–3792.

52.

Özkurt

Zerdali

. Design and implementation of hybrid adaptive extended Kalman filter for state estimation of induction motor. IEEE Trans Instrum Meas 2022; 71: 1–12.

53.

Sun

Cai

Yang

, et al. Finite position control of interior permanent magnet synchronous motors at low speed. IEEE Trans Power Electron 2022; 37: 7729–7738.

54.

Lin

C-K

Liu

T-H

J-T

, et al. Model-free predictive current control for interior permanent-magnet synchronous motor drives based on current difference detection technique. IEEE Trans Ind Electron 2014; 61: 667–681.

55.

Sun

Lei

, et al. Adaptive model-free predictive current control for SPMSM drives with optimal virtual vector modulation. IEEE ASME Trans Mechatron. Epub ahead of print 1 December 2023. DOI: 10.1109/TMECH.2023.3332136.

56.

Sun

Lei

, et al. Model-free predictive current control for dual three-phase PMSM drives with an optimal modulation pattern. IEEE Trans Ind Electron. Epub ahead of print 1 December 2023. DOI: 10.1109/TIE.2023.3327554.

57.

Carlet

Favato

Bolognani

, et al. Data-driven continuous-set predictive current control for synchronous motor drives. IEEE Trans Power Electron 2022; 37: 6637–6646.

58.

Zhang

Zhao

Cheng

, et al. Robust model predictive current control based on inductance and flux linkage extraction algorithm. IEEE Trans Veh Technol 2020; 69: 14893–14902.

59.

Zhang

Yan

Cheng

. Two-stage series model predictive torque control for PMSM drives. IEEE Trans Power Electron 2021; 36: 12910–12918.

Simplified robust two-step model predictive control for permanent magnet synchronous motor

Abstract

Keywords

Introduction

Conventional FCS-MPC of PMSMs

Mathematical model of the PMSMs

Conventional FCS-MPC strategy

Simplified robust two-step FCS-MPC

Incremental prediction model

Simplified two-step MPC

Inductance extraction algorithm

Experimental result

Steady-state performance and switching frequency comparison under normal parameters

Robust performance comparison

Conclusion

Future outlook

Footnotes

Author contributions

Declaration of conflicting interests

Funding

Institutional review board statement

Informed consent statement

ORCID iDs

Data availability statement

References