Twelve Sectors DPC Control Based on Neural Hysteresis Comparators of the DFIG Integrated to Wind Power

This may be realized by using the generator model on the fixed d-q axis set. Reactive power, active power, and rotor flux sector zone may be calculated with the help of currents and voltages measured in the generator rotor [18].

Stator active/reactive power is estimated using (1) and (2) [19].

${{P}_{s}}=-\frac{3}{2}\frac{Lm}{\sigma .Ls.Lr}\cdot (Vs.{{\psi }_{r\beta }})$     (1)

${{Q}_{s}}=-\frac{3}{2}(\frac{Vs}{\sigma .Ls}\cdot {{\psi }_{r\beta }}-\frac{Vs.Lm}{\sigma .Ls.Lr}\cdot {{\psi }_{r\alpha }})$     (2)

1.png

Figure 1. Twelve sectors DPC method

The active and reactive powers can be reformulated by inducing angle λ between the stator and rotor vectors as follows:

${{P}_{s}}=-\frac{3}{2}\frac{Lm}{\sigma .Ls.Lr}\mathop{w}_{s}\left| \mathop{\psi }_{s} \right|\left| \mathop{\psi }_{r} \right|\sin (\lambda )$     (3)

${{Q}_{s}}=-\frac{3}{2}\frac{ws}{\sigma .Ls}\left| \mathop{\psi }_{s} \right|(\frac{M}{\mathop{L}_{r}}\left| \mathop{\psi }_{r} \right|\cos (\lambda )-\left| \mathop{\psi }_{s} \right|)$     (4)

And:

$\mathop{\Psi }_{s\alpha }=\sigma \mathop{L}_{r}\mathop{I}_{r\alpha }+\frac{M}{\mathop{L}_{s}}\mathop{\Psi }_{s}$     (5)

$\mathop{\Psi }_{s\beta }=\sigma \mathop{L}_{r}\mathop{I}_{r\beta }$     (6)

$\left| \overline{\mathop{\Psi }_{s}} \right|=\frac{\left| \overline{\mathop{V}_{s}} \right|}{\mathop{w}_{s}}$     (7)

where, V_s is the stator voltage.

$\sigma =1-\frac{\mathop{M}^{2}}{\mathop{L}_{r}\mathop{L}_{s}}$     (8)

The rotor flux can be estimated by:

$\left\{ \begin{matrix}   \mathop{\Psi }_{r\alpha }=\int\limits_{0}^{t}{(}-\mathop{R}_{r}\mathop{I}_{r\alpha }+\mathop{V}_{r\alpha })dt  \\   \mathop{\Psi }_{r\beta }=\int\limits_{0}^{t}{(}-\mathop{R}_{r}\mathop{I}_{r\beta }+\mathop{V}_{r\beta })dt  \\\end{matrix} \right.$     (9)

And:

$\mathop{\psi }_{r}=\sqrt{\mathop{\Psi }_{r\alpha }^{2}+\mathop{\Psi }_{r\beta }^{2}}$     (10)

The angle of rotor flux is calculated by:

$\mathop{\theta }_{r}=arctg(\frac{\mathop{\Psi }_{r\beta }}{\mathop{\Psi }_{r\alpha }})$     (11)

In twelve sectors DPC method a 2-level HC (see Figure 2) is used for the active power (Hp) and a two-level HC (see Figure 3) for the reactive power (Hq). Finally, based on the values of constants Hq and Hp and the position of the rotor flux (12 region command), the proposed switching table is as shown in Table 1.

2.png

Figure 2. Active power hysteresis comparator

Table 1. Proposed switching table of twelve sectors DPC method

3.png

Figure 3. Reactive power hysteresis comparator

Simulations of the proposed strategies for a DFIG are conducted by using the Matlab/Simulink software. The proposed strategies will be tested and compared in two different configurations: robustness against parameter variations and reference tracking.

4.1Reference tracking test

In this section, the reactive/active power tracks almost perfectly their reference values (See Figures 13-14). On the other hand, the DPC-NHC strategy reduced the reactive and active power ripples compared to the conventional DPC strategy (See Figures 15-16).  

9.png

Figure 9. Neural network training performance

Figures 10-11 show the THD of stator current for DPC-NHC and conventional DPC method one respectively. It can be clearly observed that the THD is minimized for the DPC-NHC method (THD = 0.74 %) when compared to conventional DPC (THD = 0.32%). It is clear from the results that the DPC-NHC has satisfied performance.

10.png

Figure 10. THD (DPC)

11.png

Figure 11. THD (DPC-NHC)

12.png

Figure 12. Active power (RTT)

13.png

Figure 13. Reactive power (RTT)

14.png

Figure 14. Zoom in the active power (RTT)

15.png

Figure 15. Zoom in the reactive power (RTT)

4.2Robustness test (RT)

In this section, the parameters have been intentionally changed such as the values of the inductances L_r and L_s are divided by 2 and the resistances R_r and R_s are multiplied by 2. The DPC-NHC method reduced the active and reactive power ripples compared to the TSDPC strategy (See Figures 20-21). On the other hand, the active and stator reactive power track almost perfectly their reference values (See Figures 18-19).

16.png

Figure 16. THD (DPC)

Figures 16-17 show the THD of the current of both techniques. It can be clearly observed that the THD is minimized for DPC-NHC strategy (THD = 0.59 %) when compared to conventional DPC (THD = 1.29%). It can be concluded that the twelve sectors DPC-NHC method is more robust than the TSDPC strategy.

17.png

Figure 17. THD (DPC-NHC)

18.png

Figure 18. Active power (RT)

19.png

Figure 19. Reactive power (RT)

20.png

Figure 20. Zoom in the active power (RT)

21.png

Figure 21. Zoom in the reactive power (RT)