ERROR: LinearAlgebra.SingularException(10) with MTK v8.74+ #2412
Labels
bug
Something isn't working
Comments
|
This needs code to be reproducible. |
|
Sorry. I will debug into the model tomorrow, to locate the part of the model, which causes the error. Then I can give the code. |
|
That is the shortest example I could make of the original code. In the discrete event a external DLL is called. This affect! is left empty now. But since the error occurs during initialization, this should not matter. But of course if this model were able to run, there may occur other unrelated errors. The error disappears, if the EM model is commented out. Also if the inv model is commented out. I hope that can help. using DifferentialEquations,ModelingToolkit, OrdinaryDiffEq
using ProgressLogging
# Configuration of the simulation.
Ts=1/6000
function affect!(integ,u,p,ctx)
end
d_fun(phi,u,v,w) = (2/3*(u-v/2-w/2))*cos(phi)+sqrt(3)*2/3*(v/2-w/2)*sin(phi)
@register_symbolic d_fun(phi,u,v,w)
q_fun(phi,u,v,w) = -(2/3*(u-v/2-w/2))*sin(phi)+sqrt(3)*2/3*(v/2-w/2)*cos(phi)
@register_symbolic q_fun(phi,u,v,w)
u_fun(phi,Ug,ph)=Ug*sin(phi+pi*ph/3)
@register_symbolic u_fun(phi,Ug,ph)
hb(x) = x>0.0 ? 1e4*x*(tanh(100*(x-0.05))/2 + (0.5)) : 0.0
@register_symbolic hb(x)
ib_fun(Ib)=Ib<0 ? Ib*1e4 : Ib*1e-3
@register_symbolic ib_fun(Ib)
function inverter(; name)
@parameters Lf Rf_HSchuetz u1_inv u2_inv u3_inv Ug1 Ug2 Ug3 C Fnom Lb b_off C_ms d_b Rg Lg
@variables t u_in(t) ia(t) ib(t) ic(t) e1(t) e2(t) e3(t) idc(t) I1(t) I2(t) I3(t) Udc(t) Udc_ms(t) [irreducible=true] Id(t) Iq(t) Ud(t) Uq(t) ϕ(t) U1(t) U2(t) U3(t) Ib(t) D_b(t) [irreducible=true] I_byp(t) [irreducible=true] I_Vg1(t) I_Vg2(t) I_Vg3(t) I_Vu1(t) I_Vu2(t) I_Vu3(t) v1(t) [irreducible=true] v2(t) [irreducible=true] v3(t) [irreducible=true] v4(t) v5(t) v6(t) v7(t) v8(t) v9(t) v10(t)
dt=Differential(t)
eqs=[
I1 + I_Vg1 ~ 0,
I2 + I_Vg2 ~ 0,
I3 + I_Vg3 ~ 0,
-I_Vu1 - I_Vu2 - I_Vu3 ~ 0,
-I1 + v5/Rf_HSchuetz - v8/Rf_HSchuetz ~ 0,
-I2 + v6/Rf_HSchuetz - v9/Rf_HSchuetz ~ 0,
-I3 - v10/Rf_HSchuetz + v7/Rf_HSchuetz ~ 0,
I_Vu1 - v5/Rf_HSchuetz + v8/Rf_HSchuetz ~ 0,
I_Vu2 - v6/Rf_HSchuetz + v9/Rf_HSchuetz ~ 0,
I_Vu3 + v10/Rf_HSchuetz - v7/Rf_HSchuetz ~ 0,
-v4 + v8 ~ U1,
-v4 + v9 ~ U2,
v10 - v4 ~ U3,
v1 ~ u1_inv,
v2 ~ u2_inv,
v3 ~ u3_inv,
dt(I1)~(v1 - v5)/Lf,
dt(I2)~(v2 - v6)/Lf,
dt(I3)~(v3 - v7)/Lf,
U1~ifelse(t>3.0,u_fun(ϕ,1*Ug1,2),u_fun(ϕ,Ug1,2)),
U2~ifelse(t>3.0,u_fun(ϕ,1*Ug2,0),u_fun(ϕ,Ug2,0)),
U3~ifelse(t>3.0,u_fun(ϕ,1*Ug3,4),u_fun(ϕ,Ug3,4)),
I_byp~hb(Udc_ms-Udc),
dt(Ib)~b_off/Lb*(Udc_ms-(1-d_b)*Udc-ib_fun(Ib)),
dt(Udc)~((I_byp+Ib)*Udc_ms-1.5*Id*Ud-(0.0)*Udc/(0.02))/(Udc*C),
#dt(Udc_ms)~(-(Ib+I_byp)+idc)/(C_ms),
dt(ϕ)~2*pi*Fnom,
Id~q_fun(-ϕ,I2,I1,I3),
Iq~d_fun(-ϕ,I2,I1,I3),
Ud~q_fun(-ϕ,U2,U1,U3),
Uq~d_fun(-ϕ,U2,U1,U3),
D_b~d_b,
]
control = Ts => (affect!, [I1, I2, I3,Udc,Ib,I_byp,Udc_ms], [Lf, Rf_HSchuetz, u1_inv, u2_inv, u3_inv, Ug1, Ug2, Ug3,C,Fnom,b_off,d_b], nothing)
ODESystem(eqs, t; discrete_events = [control],continuous_events=[-Udc+Udc_ms-0.02 ~ 0],name=name)
end
@named inv=inverter()
u0inv = [ inv.I1=>0,
inv.I2=>0,
inv.I3=>0,
inv.Udc=>1000,
inv.ϕ=>0.0,
inv.Id=>0,
inv.Iq=>0,
inv.Ud=>0,
inv.Uq=>0,
inv.U1=>0,
inv.U2=>0,
inv.U3=>0,
inv.Ib=>0,
inv.Udc_ms=>1,
inv.I_byp=>0,
inv.v6=>0,
inv.v4=>0,
inv.v10=>0,
inv.v1=>0,
inv.v2=>0,
inv.v3=>0,
inv.D_b=>1,
]
pinv=[ inv.Lf=>250e-6,
inv.Rf_HSchuetz=>100,
inv.u1_inv=>0,
inv.u2_inv=>0,
inv.u3_inv=>0,
inv.Ug1=>690*sqrt(2)/sqrt(3),
inv.Ug2=>690*sqrt(2)/sqrt(3),
inv.Ug3=>690*sqrt(2)/sqrt(3),
inv.C=>15.75e-3,
inv.Fnom=>50.0,
inv.Lb=>1e-3,
inv.d_b=>1,
inv.b_off=>1,
]
h(x) = tanh(1*(x))/2 + (0.5)
@register_symbolic h(x)
f(a,b,c) = 1.224744871391589*(2*h(a)-h(b)-h(c))/3
@register_symbolic f(a,b,c)
Lμ(ihd,MdE,mode)=mode>0 ? MdE : ((ihd)./(883.2749425452234.+0.28527750033027716*(ihd)))./ihd
@register_symbolic Lμ(Ihd,MdE,mode)
@register_symbolic dq2a(id,iq,ϕ)
function dq2a(id,iq,ϕ)
iα=id*cos(ϕ)-iq*sin(ϕ)
ia=iα
return ia
end
@register_symbolic dq2b(iα,id,iq,ϕ)
function dq2b(iα,id,iq,ϕ)
iβ=id*sin(ϕ)+iq*cos(ϕ)
ib=-0.5*iα+sqrt(3)/2*iβ
return ib
end
@register_symbolic abc2d(ia,ib,ic,ϕ)
function abc2d(ia,ib,ic,ϕ)
iα=ia
iβ=(ib-ic)/sqrt(3)
id=iα*cos(ϕ)+iβ*sin(ϕ)
return id
end
@register_symbolic abc2q(ia,ib,ic,ϕ)
function abc2q(ia,ib,ic,ϕ)
iα=ia
iβ=(ib-ic)/sqrt(3)
iq=-iα*sin(ϕ)+iβ*cos(ϕ)
return iq
end
@register_symbolic phi(id,iq)
function phi(id,iq)
id_=abs(id)
iq_=abs(iq)
if id >0
if iq >= 0
phi=atan(iq_/id_)
else
phi=2*pi-atan(iq_/id_)
end
else
if iq >= 0
phi=atan(id_/iq_)+pi/2
else
phi=atan(iq_/id_)+pi
end
end
phi=isnan(phi) ? 0.0 : phi
return phi*360/(2*pi)
end
@register_symbolic mag(id,iq)
function mag(id,iq)
I=sqrt(abs(id)^2+abs(iq)^2)
return I
end
function geno(; name)
@parameters ld lq lσe r1 re ω MdE MqE ü Zp C lσa a2 a1 R sat ω0 par_gen
@variables t iout(t) id(t) iq(t) Udc(t) e1(t) e2(t) e3(t) idc(t) ie(t) Ψd(t) Ψq(t) Ψe(t) ud(t) id_(t) iq_(t) id__(t) iq__(t) uq(t) ua(t) ub(t) uc(t) ue(t) ifd(t) ufd(t) M(t) P(t) ϕi(t) ϕu(t) θ(t) ϕ(t) ia(t) ib(t) ic(t) Imag(t) ϕs(t) udc(t) ϕ_filt(t) ϕ_filtˍt(t) L_mu(t)
dt = Differential(t)
eqs = [
L_mu~Lμ(ie+id+id_,MdE,sat),
Ψd~lσa*(id+id_)+Lμ(ie+(id+id_),MdE,sat)*(ie+(id+id_)),
Ψq~lσa*(iq+iq_)+MqE*(iq+iq_),
Ψe~lσe*ie+Lμ(ie+(id+id_),MdE,sat)*(ie+(id+id_)),
dt(Ψd)~ud+ω*Ψq-r1*(id+id_),
dt(Ψq)~uq-ω*Ψd-r1*(iq+iq_),
dt(Ψe)~ue-re*ie,
ud~-id*1000,
uq~-iq*1000,
dt(id_)~ω0*(par_gen*abc2d(-ia,-ib,-ic,ϕ)-id_),
dt(iq_)~ω0*(par_gen*abc2q(-ia,-ib,-ic,ϕ)-iq_),
id__~par_gen*abc2d(-ia,-ib,-ic,ϕ),
iq__~par_gen*abc2q(-ia,-ib,-ic,ϕ),
ifd~ie*ü*1.5,
ufd~ue/ü,
M~1.5*Zp*(Lμ(ie+(id+id_),MdE,sat)*ie*(iq+iq_)+(ld-lq)*(id+id_)*(iq+iq_)),
P~M*ω/Zp,
ϕi~phi(id,iq),
ϕu~phi(ud,uq),
θ~90-ϕu,
dt(ϕ)~ω,
ua~dq2a(ud,uq,ϕ),
ub~dq2b(ua,ud,uq,ϕ),
uc~-ua-ub,
Imag~mag(id,iq),
ϕs~deg2rad(2.5),
dt(Udc)~(idc-iout)/(C),
dt(ia)~1/(lσa*par_gen)*(-e1-r1*par_gen * ia +ua),
dt(ib)~1/(lσa*par_gen)*(-e2-r1*par_gen * ib +ub),
dt(ic)~1/(lσa*par_gen)*(-e3-r1*par_gen * ic +uc),
e1~f(ia,ib,ic)*Udc,
e2~f(ib,ic,ia)*Udc,
e3~f(ic,ia,ib)*Udc,
idc~h(ia)*ia+h(ib)*ib+h(ic)*ic, ]
ODESystem(eqs;name=name,continuous_events=[ib~0,ic~0,ia~0])
end
@named HM = geno()
@named EM = geno()
u0 = [ HM.id=>0,
HM.iq=>0,
HM.ie=>10,
HM.ia=>0,
HM.ib=>0,
HM.ic=>0,
HM.id_=>0,
HM.iq_=>0,
HM.id__=>0,
HM.iq__=>0,
HM.ua=>0,
HM.ub=>0,
HM.uc=>0,
HM.e1=>0,
HM.e2=>0,
HM.e3=>0,
HM.idc=>0,
HM.Udc=>10,
HM.Imag=>0,
HM.Ψd=>0,
HM.Ψq=>0,
HM.Ψe=>0,
HM.ud=>0,
HM.uq=>0,
HM.ue=>0,
HM.ufd=>0,
HM.ifd=>0,
HM.M=>0,
HM.P=>0,
HM.ϕi=>0,
HM.ϕu=>0,
HM.ϕs=>0,
HM.θ=>0,
HM.ϕ=>0,
HM.iout=>0,
EM.iout=>0,
EM.id=>0,
EM.iq=>0,
EM.ie=>10,
EM.ia=>0,
EM.ib=>0,
EM.ic=>0,
HM.e1=>0,
HM.e2=>0,
HM.e3=>0,
EM.id_=>0,
EM.iq_=>0,
EM.id__=>0,
EM.iq__=>0,
EM.ua=>0,
EM.ub=>0,
EM.uc=>0,
EM.idc=>0,
EM.Udc=>10,
EM.Imag=>0,
EM.Ψd=>0,
EM.Ψq=>0,
EM.Ψe=>0,
EM.ud=>0,
EM.uq=>0,
EM.ue=>0,
EM.ufd=>0,
EM.ifd=>0,
EM.M=>0,
EM.P=>0,
EM.ϕi=>0,
EM.ϕu=>0,
EM.ϕs=>0,
EM.θ=>0,
EM.ϕ=>0,
]
function x_pu_in_real(Z0,f,x)
ω0=2*pi*f
return (x*Z0)/ω0
end
Z0=0.163
xhd=1.913
xhq=1.041
xd=2.035
xq=1.163
xσa=0.1218
ra=5.82e-3
xσf=0.2802
rσf=1.47e-3
ü_=sqrt(2/3*(rσf*Z0)/0.81220)
Z0_EM=0.502
xhd_EM=1.07
xhq_EM=0.358
xd_EM=1.247
xq_EM=0.531
xσa_EM=0.1729
ra_EM=26.15e-3
xσf_EM=0.2106
rσf_EM=4.6e-3
ü_EM=sqrt(2/3*(rσf_EM*Z0_EM)/3.20863309352518)
ω_0=2*pi*1e4
D=100
p = [ HM.r1=>ra*Z0,
#-----------------------
HM.re=>rσf*Z0,
HM.lσe=>x_pu_in_real(Z0,68.1,xσf),
HM.MdE=>x_pu_in_real(Z0,68.1,xhd),
HM.MqE=>x_pu_in_real(Z0,68.1,xhq),
HM.lσa=>x_pu_in_real(Z0,68.1,xσa),
#-----------------------
HM.ld=>x_pu_in_real(Z0,68.1,xd),
HM.lq=>x_pu_in_real(Z0,68.1,xq),
#-----------------------
HM.ω=>pi*68.1,
HM.ü=>ü_,
HM.Zp=>6,
HM.C=>4.5e-3,
HM.sat=>0,
HM.ω0=>2*pi*5,
HM.par_gen=>6,
EM.par_gen=>1,
EM.r1=>ra_EM*Z0_EM,
#-----------------------
EM.re=>rσf_EM*Z0_EM,
EM.lσe=>x_pu_in_real(Z0_EM,136.2,xσf_EM),
EM.MdE=>x_pu_in_real(Z0_EM,136.2,xhd_EM),
EM.MqE=>x_pu_in_real(Z0_EM,136.2,xhq_EM),
EM.lσa=>x_pu_in_real(Z0_EM,136.2,xσa_EM),
#-----------------------
EM.ld=>x_pu_in_real(Z0_EM,136.2,xd_EM),
EM.lq=>x_pu_in_real(Z0_EM,136.2,xq_EM),
#-----------------------
EM.ω=>2*pi*68.1,
EM.ü=>ü_EM,
EM.Zp=>12,
EM.C=>4.5e-5,
EM.sat=>1,
EM.ω0=>2*pi*10,
]
append!(u0,u0inv)
append!(p,pinv)
tspan = (0.0, 30)
@variables t
connected = compose(ODESystem([
HM.ufd~10,
HM.iout~inv.Ib+inv.I_byp,
inv.Udc_ms~HM.Udc,
EM.iout~HM.ifd,
EM.ufd~5.0,
], t; name = :connected), EM,HM,inv)
oprob = ODEProblem(structural_simplify(connected), u0, tspan, p)
sol = solve(oprob, Rodas4P2(),reltol=1e-3,maxiters = 1e7, progress = true,saveat=tspan[2]*1e3)#,callback=downcrossing_cb)
|
|
I think we should hold off on this until the v9 release since that will have new initialization features which may fix this. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
With a model running fine with MTK v8.73.2 i run into an error with MTK v8.74.0 and also the latest v8.75. The stacktrace is attached. Since the model is little bit complex it's difficult to provide example code. Same error with Julia v1.9.2 and v1.10.
The text was updated successfully, but these errors were encountered: