On Wed, Aug 8, 2018 at 12:10 PM, Beginner1 <address@hidden> wrote:
Hi,
There is an error in how Octave first simplifies and then multiplies
matrixes.
Here is my code:
/*pkg load control
pkg load io
pkg load signal
%eig_A=eig(sys.A);
%plot(real(eig_A),imag(eig_A),"rx")
%grid on
%% se define la variable para hacer el barrido
%cycle_vector = logspace(log10(0.5),log10(2),100);
cycle_vector = linspace(0,15,15);
color_plot = colormap(jet(length(cycle_vector))); %los colores van variando
del azul al rojo
%all = axes('position',[.1,.1,.8,.8])
%% cycle (secalcula las nuevas matrices para cada valor del barrido)
for n_cycle = 1:length(cycle_vector)
%hay que recalcular la matrz A para el barridod e la variable puede que
haya que Volver a linealizar
kv=cycle_vector(n_cycle);
%kv=15;
Tv=0.05;%1;%Constante de tiempo del PI de la componente d
kq=0.4;%Constante proporcional del PI de la componente q
Tq=0.02;%Constante de tiempo del PI de la componente q
kdroop_AC=0;%-3;%Pendiente droop Q/U_AC
droop_FSM=0.05;%Pendiente droop P/f
FSM_slope=droop_FSM/0.002;%Pendiente droop P/f en p.u.
%Inner Loop
ki=5000;%1% Constante integral de los controladores proporcional
resonantes
kp=1.5;% Constante proporcional de los controladores proporcional
resonantes
w=2*pi*50;
wg0=2*pi*50.01;
w_r=2*pi*50;
wr=2*pi*50;
a23 = 2*(wg0^2+wr^2);
da23 = 4*wg0;
a03 = (wr^2-wg0^2)^2;
da03 = -4*wg0*(wr^2-wg0^2);
b33 = ki;
db33 = 0;
b13 = ki*(wg0^2+wr^2);
db13 = ki*2*wg0;
c23 = -ki*wg0;
dc23 = -ki;
c03 = ki*wg0*(wr^2-wg0^2);
dc03 = ki*(wr^2-3*wg0^2);
%2-level VSC&Normalization block
Vdref_perm=1.002;%1.003;
Vqref_perm=0.085;%0.077;
Vdc_diff_perm=0.5;%0.492;
kdiff=(0.7)*Vdc_diff_perm;%Según DIgilent debería ser multiplicada por 1.4
en vez de 0.7
K_0=sqrt(3)/(2*sqrt(2));%Factor de modulación sinusoidal del VSC
Ub_DC=640/2;%Tensión base en DC del VSC-Según DIgilent debería ser 640 en
vez de 640/2
Ub_AC=275;%Tensión base en AC del VSC
k0=(K_0)*Ub_DC/Ub_AC;
Vdc_perm=1.001;%1.017;
Pmd_perm=0.702;%0.702;%0.69;
Pmq_perm=0;%0.053;
Kvsc=k0*Vdc_perm;
x1d_0= 0;
x1q_0= 0;
x2d_0=0;
x2q_0=0;
x3d_0=0;
x3q_0=0;
x4d_0=0;
x4q_0=0;
%Función enable: para marcar la activación de la rama FSM
enable=0;
%Definición de matrices A,B,C y D
%Outer loop:Vdc-Q
A_ol=[0 0;0 0];
B_ol=[-kv/Tv kv/Tv 0 0 0 0 -kv*FSM_slope*enable/Tv kv*enable/Tv
-kv*enable/Tv kv*enable/Tv;0 0 kq/Tq -kq/Tq -(kq/Tq)*kdroop_AC
(kq/Tq)*kdroop_AC 0 0 0 0];
C_ol=[1 0;0 1];
D_ol=[-kv kv 0 0 0 0 -kv*FSM_slope*enable kv*enable -kv*enable kv*enable;0
0 kq -kq -kq*kdroop_AC kq*kdroop_AC 0 0 0 0];
%Inner loop:PR current controller
A_il=[0 1 0 0 0 0 0 0; 0 0 1 0 0 0 0 0;0 0 0 1 0 0 0 0; -(w^2-wg0^2) 0
-4*w^2 0 0 0 0 0;0 0 0 0 0 1 0 0;0 0 0 0 0 0 1 0;0 0 0 0 0 0 0 1;0 0 0 0
-(w^2-wg0^2) 0 -4*w^2 0];
B_il=[0 0 0 0 0 0 0;0 0 0 0 0 0 0;0 0 0 0 0 0 0;0 0 -1 0 1 0 -4*w*x3d_0;0
0 0 0 0 0 0;0 0 0 0 0 0 0;0 0 0 0 0 0 0;0 0 0 -1 0 1 -4*w*x3q_0];
C_il=[0 2*ki*w^2 0 ki 0 0 ki*w 0; 0 0 -ki*w 0 0 2*ki*w^2 0 ki];
D_il=[1 0 -kp 0 kp 0 ki*(2*w*x2d_0+x3q_0+2*w^2*x1q_0); 0 1 0 -kp 0 kp
ki*(2*w*x2q_0-2*w^2*x1d_0-x3d_0)];
%Normalization block
D_nb=[kdiff 0 (0.7)*Vdref_perm;0 kdiff (0.7)*Vqref_perm];%Según DIgilent
debería ser multiplicada por 1.4 en vez de 0.7
%2-level VSC:controlled voltage source-average model
D_vsc=[k0*Vdc_perm 0 k0*Pmd_perm;0 k0*Vdc_perm k0*Pmq_perm];
%Outer loop
stname_ol = {'xv','xq'};
inname_ol={'vdcref','vdc','qacref','qac','vacref','vac',' fmeas','pdcref','paux',' FSMcoordpos'};
outname_ol={'idref','iqref'};
sys_ol = ss
(A_ol,B_ol,C_ol,D_ol,'stname',stname_ol,'inname',inname_ol,' outname',outname_ol);
%Inner loop
stname_il = {'xd1','xd2','xd3','xd4','xq1','xq2','xq3','xq4'};
inname_il={'Vd','Vq','Id','Iq','idref','iqref','omegagiro'};
outname_il={'Vdref','Vqref'};
sys_il = ss
(A_il,B_il,C_il,D_il,'stname',stname_il,'inname',inname_il,' outname',outname_il);
%Normalization block
inname_nb={'Vdref','Vqref','Vdcdiff'};
%inname_nb={'Vdref','Vqref'};
outname_nb={'Pmd','Pmq'};
sys_nb = ss (D_nb,'inname',inname_nb,'outname',outname_nb);
%2-level VSC
inname_vsc={'Pmd','Pmq','vdc'};
outname_vsc={'Vd_AC','Vq_AC'};
sys_vsc = ss (D_vsc,'inname',inname_vsc,'outname',outname_vsc);
%Assembled system1
inname_group={'vdcref','vdc','qacref','qac','vacref','vac',' fmeas','pdcref','paux',' FSMcoordpos','Vd','Vq','Id',' Iq','omegagiro','Vdcdiff'};
%inname_group={'vdcref','vdc','qacref','qac','vacref','vac', 'fmeas','pdcref','paux',' FSMcoordpos','Vd','Vq','Id',' Iq'};
outname_group={'idref','iqref','Vdref','Vqref','Pmd','Pmq',' Vd_AC','Vq_AC'};
sys=connect(sys_ol,sys_il,sys_nb,sys_vsc,inname_group, outname_group);
B=sys.B;
K_cl=ones(16,9);
Aol=sys.A;
Acl=Aol+B*K_cl;
end*/
You can copy and paste this code and see what hapens afterwads. The error I
get is
error: rrr: operator +: nonconformant arguments (op1 is 10x10, op2 is 10x9)
error: called from
rrr at line 131 column 6
But what is important in this message is the last part of the code, that
refers to the sum of matrixes:
/* B=sys.B;
K_cl=ones(16,9);
Aol=sys.A;
Acl=Aol+B*K_cl;*/
The dimension problem should be solved by making K_cl equal to ones(16,10).
But when I do this, and run the code, the console reports again a new
dimension problem:
error: rrr: operator +: nonconformant arguments (op1 is 9x9, op2 is 9x10)
error: called from
rrr at line 131 column 6
So Octave must be simplifying matrixes since they are plenty of zeros and
this does not allow to perform the sum of them as they cannot coincide in
dimension.
Any suggestion to avoid the simplification of matrixes in Octave?
--
Sent from: http://octave.1599824.n4.nabble.com/Octave-General- f1599825.html
I modified the end of your file to: