The specific harmonic elimination modulation method was proposed by Patel H.S. and Hoft R.G. in 1973, and it is mainly used in the field of variable frequency power supply [
18,
19]. When the three-level inverter is used in the case of high-power motor drive, in order to reduce the switching frequency of power devices and reduce switching losses, specific harmonic elimination technology (SHET) is needed to eliminate the low-order harmonics in the input voltage waveform. Therefore, the current ripple can be reduced, and the performance of the variable frequency power supply can be improved [
20,
21].
The main principle of SHET is based on the Fourier series model of the input phase voltage of three-phase inverter. The specific subharmonic amplitude to be eliminated is set to zero as the constraint, and the solution is based on the switching angle α nonlinear transcendental equations of unknown quantities. Finally, the specific harmonic in the inverter input voltage is eliminated through modulation.
2.1. Principle of Magnetic Pole-Specific Harmonic Elimination
In this paper, the above SHET was applied to the design of surface-mounted rotor magnetic poles in order to reduce the harmonics of the air-gap flux density in the motor and reduce the torque ripple. The specific harmonic in the no-load air-gap flux density can be eliminated by solving the width and position angle of the modulated magnetic pole through the SHET. Only the sinusoidal fundamental wave was retained to realize the sinusoidal air-gap flux density of the motor. In actual calculations, a symmetrical arrangement is usually adopted to simplify the calculation. The theoretical waveforms of modulated and sinusoidal magnetic poles are shown in
Figure 2. Additionally, a cross-section picture of the machine in the position of different
αim is shown in
Figure 3.
As shown in
Figure 2, the waveform is symmetrical in a period of positive and negative half periods. The waveform in the positive half period is symmetric at about
π/2, where
αi (
i = 1, 2, 3, …,
N) is the position angle of each magnetic pole in a quarter period, and
αim is the corresponding mechanical angle. If there are
N position angles in a quarter period (
N − 1), specific harmonics can be eliminated. Under a pair of poles, the Fourier series expansion expression of the profile waveform of the modulated magnetic pole is as follows:
where
n is the order of each harmonic;
cn and
dn are the amplitudes of the sine and cosine components, respectively; and the expressions are as follows:
Since the magnetic pole positions are oddly symmetric at around the point π and symmetric at about
π/2 in the interval [0,
π], it follows that
Substituting Equations (4) and (5) into Equations (2) and (3), respectively, we can obtain the following:
where
h is the magnetic pole height and
αi denotes the
i-th position angle. From Equation (6), it can be seen that all even harmonics in the modulated pole profile waveform can be eliminated, and all cosine harmonics are also eliminated. Therefore, the amplitude of each harmonic in the modulated pole profile waveform can be expressed as follows:
where the position angle
αi should satisfy 0 <
α1 <
α2 < α
3 … <
αN <
π/2. In the magnetic field generated by the magnetic poles, the effects of the 3rd harmonic and its multiple harmonics in the air-gap flux density need to be considered, so the harmonics to be eliminated are 3, 5, 7, 9, 11, …, 2
k − 1 (
k = 2, 3, 4, …). In this paper, a 12-slot 10-pole surface-mounted dual three-phase permanent magnet motor was designed as an example. The harmonic distribution of air-gap flux density generated by the conventional surface-mounted magnetic pole structure is shown in the following figure.
It can be seen from
Figure 4 that 3, 5, 7 and 9 non-working harmonics with high amplitudes are the main factors affecting the sinusoidality of the air-gap magnetic density. The number of position angles
N = 5 in a quarter period is chosen, so that four specific harmonics of the 3rd, 5th, 7th and 9th can be eliminated. Define
bn as the
n-th harmonic amplitude, and define the modulation index
M as the ratio of the square wave fundamental amplitude
b1 to the square wave amplitude
hm, i.e.,
The modulation index is
M = 1, and the nonlinear equations composed of the corresponding harmonic amplitude expressions are as follows:
Therefore, the position angle of each magnetic pole can be obtained only by solving the solution of the equations. The magnetic pole width is further obtained, followed by the elimination of specific harmonics in the air-gap magnetic density. However, this system of nonlinear transcendental equations cannot be solved directly, and the initial value of αi needs to be calculated and solved using an iterative method.
The solution of the nonlinear system of equations is the core of obtaining the width and position of the modulated magnetic pole. This initial value is selected using the principle of impulse equivalency. As shown in
Figure 3 and
Figure 5, there is one pole for each half period. Additionally, the pole is divided into
P rectangular segments with different widths. The center of each pole is equidistant, the spacing is
π/
P and the amplitude is
hm. Thus, the centerline position angle of the
i-th pole is as follows:
The height of the sine waveform corresponding to the center position of the
i-th magnetic pole is as follows:
The width
Wi of the
i-th magnetic pole can be calculated:
Therefore, the starting and ending position angles of the
ith magnetic pole are as follows:
Matlab has efficient numerical and symbolic computation functions, and it contains a large number of computational algorithms. The Matlab fslove function included in its toolbox can directly solve large systems of nonlinear transcendental equations, and its basic iteration principle is the Newton iteration method. The initial value of
αi is used as the initial value of the fslove function of the iterative solution. Additionally, the correct setting of the initial value of
αi is conducive to the convergence of the program. When the program converges, the approximate true solution of nonlinear transcendental equations can be obtained. The magnetic pole position parameters obtained from the actual solution are shown in
Table 1.
The relationship between the mechanical position angle
αim and the electric angle
αi of the modulation pole position is given as follows:
the mechanical position angle
αim of the modulated magnetic poles under one pole can be obtained as shown in
Table 2.
Therefore, according to symmetry, the width of five modulated magnetic poles under one pole can be obtained:
In this paper, the rotor inner radius
R is 22.5 mm, and the final width of each magnetic pole under one pole is shown in
Table 3.
2.2. Principle of Magnetic Pole-Specific Harmonic Injection
In some occasions where the performance of the motor is required to be high, in addition to requiring the motor to have low torque ripple to enable the motor to operate smoothly, it also requires the motor to have high torque density. As we all know, when the surface-mounted motor rotor magnetic pole is cut into an extremely sinusoidal structure, although the torque ripple will be reduced, the torque density will also be greatly reduced. Therefore, in order to increase the torque without aggravating the torque ripple, the solution of injecting the third harmonic into the sinusoidal pole shape is proposed. The fundamental air-gap magnetic density amplitude can be increased by injecting the third harmonic while keeping the maximum magnetic pole thickness constant. Not only that, in the case of a dual three-phase motor with two sets of windings shifted 30° in phase, the torque density can be further increased by injecting the third harmonic current [
22], which interacts with the third harmonic in the air-gap magnetic density to generate electromagnetic torque.
After injecting the third harmonic into the sinusoidal magnetic pole shape, the pole shape function at this point is as follows:
where Δ
ma is the adjustment factor whose magnitude varies with the change in the harmonic injection amount
a. This ensures that the maximum thickness of the PM before and after harmonic injection is kept unchanged without changing the minimum air-gap length.
a is the amplitude of the third harmonic injection. The theoretical waveforms of the 3rd harmonic injecting sin waveform are shown in
Figure 6. Neglecting the effect of slot opening and assuming the same relative permeability of air gap and PM, the air-gap magnetic density of the dual three-phase PMSM can be expressed as follows:
where
Br is the remanence of the PM, and
l is the air-gap length. Each back electromotive force of the motor can be derived from the magnetic flux linkage equation:
where
Rs is the stator inner diameter;
la is the axial length of stator core;
ωr is the rotor angular speed;
Np is the number of series turns per phase; and
Kdpn is the winding coefficient of each harmonic. By passing a sinusoidal current through the dual three-phase winding, the electromagnetic torque generated by each back electromotive force can be expressed as follows:
After the third harmonic is injected into the sinusoidal pole, in order to solve for the harmonic injection amount a at the maximum PM thickness, the differential Equation (17) can be obtained:
Therefore, the maximum thickness of PM is as follows:
At this time, the maximum value of the magnetic pole shape function
h1o (
θ) is as follows:
After the third harmonic is injected into the sinusoidal magnetic pole, to keep the maximum thickness of the PM constant before and after injection, let
h1omax =
hm; thus:
The output torque of the motor is obtained as follows:
When
a = 1/9, the torque after the third harmonic injection reaches the maximum; when
a > 1/9, the optimal third harmonic injection amount
a is solved using the Lagrangian function. First, the Lagrangian function is established:
where
λ is the Lagrangian coefficient; thus, the maximum torque of the dual three-phase PMSM provided by the third harmonic injection into the sinusoidal pole can be determined by differentiating the equation
L(∆
ma,
a) as follows:
The simultaneous solution of Equation (27) can be obtained:
Therefore, in order to ensure the same maximum magnetic pole thickness before and after injection and to obtain the maximum torque at the same time, the optimal injection amount for the third harmonic injection into the sinusoidal pole is 1/6. At this point, the magnetic pole shape function is as shown in Equation (29), and the magnetic pole shape after the injection of the third harmonic is shown in
Figure 7.
As can be seen from
Figure 7, the magnetic pole shape function contains only fundamental and third harmonics, and the equation group composed of the amplitude expression of each harmonic is as follows:
Using Matlab fslove function to solve (30), the position angle parameters of equivalent sin + 3rd PM structure are obtained as shown in
Table 4,
Table 5 and
Table 6. According to the symmetry of the waveform with respect to
π/2 in the positive half period, the corresponding mechanical position angle
αim of the modulated magnetic poles under one pole and widths of the five magnetic poles can be obtained as follows.