Professional Documents
Culture Documents
The mathematical model of PMSM based on SVPWM with PI consists of two dynamics. They are:
simulation of model predictive control with SVPWM based PMSM using ii) Mechanical dynamics.
Simulink block is peaceful when compared to m-code is explained in [5]. The general block diagram of PMSM model in the rotating
The Model Predictive Control (MPC) is to a certain degree frame (dq-axis) are shown in the Fig. 1.
progress compared the previously explained ideal flux path technique. The A. Electrical dynamics
researchers did many experiments in the PMSM motor with MPC controller The differential equations of PMSM model with respect to
in a different point of views. Among them, some of the few transactions are electrical dynamics part can be expressed as
1
By Park’s transformation,
The 2ϕ Stationary co-ordinates are transformed into
synchronous rotating co-ordinates. These expressions are as equ. 8,
[ ] * +[ ] (8)
Fig. 1.General Block Diagram of PMSM Model – Rotating Frame generated and it given as the input to proportional plus integral (PI-1)
controller. There are 3 PI controller is involved in the SVPWM control
based PMSM and is shown in the Fig. 3.
The PI-1 gives the value of isqref & isq from the park
transformation block is given to the comparator & error signal produced
(1) and then the signal set to the PI-2 input to get the Vsqref. Similarly, isdref and
The electromagnetic torque developed by PMSM rotor can be isd from the park transformation block is compared and signal generated is
[ ( ) ] (2)
B. Mechanical dynamics
The differential equations of PMSM model with respect to
mechanical dynamics part can be expressed as
( (3)
( ( ) ( ( ( ( (4)
basic technique to take the ideal flux path, when the PMSM is powered by Duty Cycle, D = 0.5 * (1 + [dx, dy, d0])
3ϕ sine wave voltage [6]. By approaching the flux circle with the actual The Vsqref & Vsdref is given as the input the inverse park transformation
flux made by inverter, so motor could acquire constant sine magnetic field. which is transformed from d-q axis to αβ components. The Vsαref & Vsβref are
By the way it is easy to achieve high performance with reduced torque agreed to generate the gate pulses in the SVPWM. The six gate pulses is
ripple. generated from the SVPWM block is set as the input the 3ϕ inverter which
In the process of controlling PMSM, the actual current in is coupled with the 3ϕ PMSM motor.
armature winding in 3ϕ ac current ia, ib & ic is undergone certain The Clark transformation is involved to produce the two αβ
transformation should be done to get current components of d-q axis. components from the abc voltage produced in the inverter output side as per
the equation (7). The cycle is fully follows the closed path to produce the
By Clark’s transformation,
sine magnetic field and produce continuously.
The 3ϕ Stationary co-ordinates are transformed into 2ϕ The same principle is simulated in MATLAB as per the
Stationary co-ordinates. These expressions are as equ.7, numerical calculation made from the theoretical explanation and simulated
diagram and its results are clearly presented in the following section. The
sector number determination and its arguments values calculated to find the
[ ] √ [ √ √
][ ] (7)
duty cycle is tabulated in Table 2.
2
present in the electromagnetic torque output. Because current is
proportional to the torque in the output side shown in the fig. 5(e).
(b)
b. Simulation of SVPWM Control
The simulation of SVPWM control based PMSM consists of the
following blocks. They are i) Reference Speed, ii) Proportional-Plus-
Integral (PI) controllers, iii) Comparators, iv) Feedback block, v) SVPWM,
vi) Park’s Transformation, vii) Clark transformation, viii) U-bridge
Converter, ix) Dc voltage source, x) PMSM, xi) Derivative w.r.to time, xii)
Display (Scope).
(c)
3
Also, there is a chance to produce the acoustic noise because of Vq (x,y) = M(x) sin[θv (x,y) - θr] (12)
the harmonics current and torque ripple present is 9.375%. The rotor speed Where θr ∊ [0,2π] is the rotor flux angle.
and torque waveform is shown in the fig. (c) & (d). The electromagnetic
torque produced in the motor is 5.2Nm. The speed settling time in the
controller based motor is 1.56s ≈ 1.6s.
where,
esd (k )= - ( Lqisq(k) (15)
esq (k )= - ( [Lqisq(k)+ ψPM] (16)
isdxy(k+1) & isqxy(k+1) are the predictions of isd & isq w.r.to vs =
dxyVxy at instant (k+1)Tc. isd(k) & isq(k) are the sampling quantities of isd and
isq at instant kTc.
Fig.6. Block diagram of MPC Control based PMSM The advantages of MPC control are i)compactness, ii) without
The earlier step of the MPC control is the compares the cascaded duty cycle calculations, iii) efficiency by the way of simplified
reference speed wm* with the actual speed wm and error signal generates is evaluations, iv) good performance in the lower ripples and harmonics
given as a input the PI controller. The output of the PI controller is disortion in the steady-state operation.
separated into the two signals; one is directly taken as torque reference and
other signal is fed into the Maximum Torque per Ampere (MTPA) gives the
flux reference. These signals are directly given to the MPC controller which
consists of cost function minimization, torque & flux prediction and
estimation. The voltage source vs get from the MPC block and send it to the
SVPWM produces pulses for the voltage source inverter to the PMSM
motor. These blocks are shown in the fig. 6. And the MPC controller is
designed on the basis of the following expressions (9)-(16).
b. Simulation of MPC Control
The MPC controls based PMSM consist of the following
fragments are dq-abc, PWM inverter, PMSM, MPC are shown in the fig.7. (a)
The dq-abc gets the id*, iq*,i0, theta taken as feedback from the PMSM
motor output. The current iabc* is engaged to PWM inverter is set to PMSM
motor. The MPC controller acquires duty cycle and angle output vector
simultaneously to solving the optimization problem.
To launch the rotating d-q coordinates & support the rotor flux
on the d-axis. The stator voltage equation of PMSM in the rotating d-q
coordinates can be expressed as:
dxyVd (x,y) = Rsisd + Ld isd - Lqisq (9)
Ld≤Lq and Vd & Vq are the d-axis and q-axis projections of vector Vxy,
which can be expressed as:
Vd (x,y) = M(x) cos[θv (x,y) - θr] (11)
4
and maintains the ideal flux path by choosing the vector in svpwm with
DBPC. The block diagram of DBPC based pmsm is shown in the fig. 9.
(c)
Fig.9. Block diagram of DBPC Control based PMSM
b. Simulation of DBPC Control
The Dead-Beat (DB) controller designed on the basis of the
following equations with delay compensation techniques are used for quick
response. The succeeding fig.10. Shows the Simulink bock used inside the
Deadbeat controller with the data store memory from the feedback of the
motor.
The current prediction vector is calculated as
Ḯ(k+1) = ik + T{ Fkik + dk + L0-1vk – (1/T)(Kηηk) } (17)
(d)
With Ḯk = i0 where Kη ∊ 𝕽2x2 symmetric gain matrix, &
ηk = ∑ (18)
The prediction error can be defined by,
ek = ik - Ḯk (19)
(e)
Fig. 8. Simulation Results obtained from MPC based PMSM Control
(a) Input DC Voltage Vdc, (b) Phase Output Currents iabc, (c) Rotor
Speed ωm, (d) Electromagnetic torque developed, (e) Ripple Content in
the output torque Fig.10. Dead-Beat Predictive Controller Simulink block
From the simulation results, Vdc is given from a dc source of The DBPC block is composed of required speed reference ωm*, d-q
300v is mutual for all the three investigators which is shown in the fig. 8(a). axis reference or prediction current idq*, actual speed ωm as feedback, 3ϕ
The output phase currents produces sine waveform of 9A with distorted stationary currents iabc from the motor, angular rotational theta θr are given
harmonics in fig. 8(b). Due to the harmonics present in the current as inputs to produce the voltage vαβ for the SVPWM block input set value.
waveform is moderately, less ripple present in the electromagnetic torque The triggering pulse also required for the block of sampling frequency 5
output. Because current is proportional to the torque in the output side kHz.
shown in the fig. 8(e). Also, there is a chance to produce the acoustic noise
because of the harmonics current and torque ripple present is 6.55%. The
rotor speed and torque waveform is shown in the fig. 8(c) & 8(d). The
electromagnetic torque produced in the motor is 8.2Nm. The speed settling
time in the controller based motor is 0.0066s.
(a)
(f)
Fig. 12. Simulation Results obtained from DBPC based PMSM Control
(a) Input DC Voltage Vdc, (b) Phase Output Currents iabc, (c) Rotor
Speed ωm, (d) Electromagnetic torque developed, (e) Ripple Content in
the output torque, (f) Rotor angle & torque vs time
The rotor speed and torque waveform is shown in the fig. 12(c)
& 12(d). The electromagnetic torque produced in the motor is 12.2Nm. The
speed settling time also less in the controller based motor is 0.005s.
(b)
(d)
VIII. Conclusion
A substantial analysis taken over with PMSM drive under
SVPWM, MPC & DB with delay compensation control is carried out in this
(c) (j)
paper. The analysis is fully based on the reduction of torque ripple,
harmonics and acoustic noise production through numerically, simulations
and implementation in hardware in the laboratory. Among all the
investigator’s, DBPC performs in the fast response, harmonic reduction in
the phase currents, torque ripple (≈2%) provides better performance results
(d) (k)
both the simulation as well as hardware testing. The compactness, good
performance & efficiency have shown from the comparative numerical
simulations & experimental studies.
List of Nomenclature
(e) (l)
Stator Current Components w.r.to Direct Axis
Stator Current Components w.r.to Quadrature Axis
Stator Voltage Components w.r.to Direct Axis
Stator Voltage Components w.r.to Quadrature Axis
w.r.to With respect to
Stator Inductance Components w.r.to d & q Axis
(f) (m) Number of Pole Pairs
Mechanical Speed w.r.to Rotor frame
Electromotive Force(EMF) Constant
Electro-Magnetic Torque developed in PMSM
Load Torque
J Moment of Inertia
Sampling Period
ia,ib,ic Actual armature current components
(g) (n)
id,iq Stationary current components referred to dq axis
iα, iβ Two stationary current co-ordinates
Fig. 14. Experimental Waveforms with different dc input voltages Align Rotor flux on d-axis
Permanent Magnet Flux Modulus
vdc= 300v (a-g) & vdc= 200v (h-n)
Rs Stator resistance
Rotor flux angular velocity
Table 3. Parameters of the Employed PMSM Drive isq & isd d-axis & q-axis projections of Stator Current is
Tc Control Period
Symbol Description Value arg Argument Values
P Rated Power 10.7 kW mag Magnitude Values
Rs Stator Resistance 1.3Ω (ohm) MPC Model Predictive Controller
Ls Stator Inductance 0.000835H MTPA Maximum Torque per Ampere
PM Rotor Magnetic Flux 0.175wb DBPC Dead Beat Predictive Controller
B Viscous friction Coefficient 0.001 kg m2 s-1 PMSM Permanent Magnet Synchronous Motor
p No. of Pole Pairs 4 SVPWM Space Vector Pulse Width Modulation
J Moment of Inertia 0.0008 kg m2
REFERENCES
[1] R Reshma, Vishnu J, Bijimol P S, “Electric Vehicle Control using
PMSM”, International Journal of Innovative Research in Electrical,
Electronics, Instrumentation and Control Engineering, Vol. 1, Special Issue
2, March 2018.
7
[2] Srinivas Pratapgiri_ , Prasad Polaki Venkata Narsimha,” Direct torque [19] T. Nguyen-Van and N. Hori, “Discretization of non-autonomous non-
control of 4 phase 8/6 switched reluctance motor drive for constant torque linear systems based on continualization of an exact discrete-time model,”
load”, World Journal of Modelling and Simulation, ISSN 1 746-7233, Vol. 8 J.Dynamic Syst., Meas. Control, vol. 136, no.2, p.021004, Nov.2013.
(2012) No. 3, pp. 185-193.
[10] Yan Yan, Shuai Wang, Changliang Xia, Huimin Wang, and Tingna
Shi, “Hybrid Control Set-Model Predictive Control for Field-Oriented
Control of VSI-PMSM”, IEEE Transactions on Energy Conversion, Vol.
31, No. 4, Pg.No. 1622-1633 December 2016.
[11] Andrés Mora, Álvaro Orellana, Jorge Juliet, and Roberto Cárdenas,”
Model Predictive Torque Control for Torque Ripple Compensation in
Variable-Speed PMSMs”, IEEE Transactions on Industrial Electronics,
Vol. 63, No. 7, Pg.No. 4584-4592, July 2016.
[12] Zbynek Mynar, Libor Vesely, and Pavel Vaclavek, “PMSM Model
Predictive Control With Field-Weakening Implementation”, IEEE
Transactions on Industrial Electronics, Vol. 63, No. 8, Pg.No. 5156-5166,
August 2016.
[14] Kvˇ etoslav Belda and David Vošmik, “Explicit Generalized Predictive
Control of Speed and Position of PMSM Drives”, IEEE Transactions On
Industrial Electronics, Vol. 63, No. 6,Pg.No. 3889- 3896, June 2016.
[17] Jesper Moos, “Predictive Deadbeat Control For PMSM Drive”, in the
submission of Master's Thesis - Power Electronics and rives, Department
of Energy Technology, Aalborg University, Denmark, June 2014.