📄 lpvchkpq.m
字号:
function [eigx,eigy,eigxy,fail] = ... lpvchkpq(vlpv,dim,vnu,fgparm,gradfg,psi,gam,xmat,ymat)% [eigx,eigy,eigxy,fail] = % lpvchkpq(vlpv,dim,vnu,fgparm,gradfg,psi,gam,xmat,ymat)% checks LMI solutions to the parameter-dependent OSF LPV control problem% with dynamic parameter measurement.%% INPUTS: % vlpv: (VARYING) matrix containing L evaluations of the OLIC:% vlpv = vpck([olic1;...;olicL],1:L)% Any direct state measurements must be ordered last (in A and in Cy).% dim = [nmeas nctrl NX NY nsf]: # of measurements, controls, X basis functions,% Y basis functions, and (optionally) directly measured states.% vnu: Np*Nq-point VARYING (2s x nbnds) matrix grid of parameter velocities% and pdot-q.% nbnds = 1: -vnu(i) < parmv(i) < vnu(i)% -vnu(s+i) < parmv(i)-q < vnu(s+i)% nbnds = 2: -vnu(i,1) < parmv(i) < vnu(i,2)% -vnu(s+i,1) < parmv(i)-q < vnu(s+i,2)% nbnds > 2: vnu(1:s,j) is the j-th corner of the velocity polytope.% vnu(s+(1:s),j) is the j-th corner of the pdot-q polytope.% fgparm=abv(fparm,gparm): fparm,gparm are basis function data for X(p,q),Y(p,q)% gradfg=abv(gradf,gradg): gradf,gradg are basis gradient data for X(p,q),Y(p,q)% VARYING matrices with Np*Nq points (p = outer loop, q = inner loop)% sel(fparm,i,1) = f_i(p,q),% sel(gradf,i,j) = df_i(p,q)/dp_j for j=1,...,s% sel(gradf,i,j) = df_i(p,q)/dq_j for j=s+1,...,2s% psi: (s x 1) vector of bandwidths for parameter measurement filters% gam optimal gamma from solver% xmat,ymat varying-matrix LMI solutions from solver%% OUTPUTS: % eigx max eigenvalues of output injection matrices (X)% eigy max eigenvalues of state feedback matrices (Y)% eigxy min eigenvalues of spectral radius matrices (XY)% fail 3-column matrix of indices to LMIs that failed% (column 1 = grid point #, column 2 = vertex # of rate polytope,% column 3 = 0/1/2 for XY/X/Y)if nargin == 0 disp('[eigx,eigy,eigxy,fail] = '); disp('lpvchkpq(vlpv,dim,vnu,fgparm,gradfg,psi,gam,xmat,ymat)'); returnend% Get some problem dimensionsnmeas = dim(1); nctrl = dim(2); NX = dim(3); NY = dim(4);if length(dim) > 4, nsf = dim(5); else nsf = 0; endgam = vunpck(gam);ngam = length(gam);% If vlpv, xmat, or ymat aren't varying matrices, vpck them[typ,dum2,dum3,npts] = minfo(vlpv);if typ == 'syst' vlpv = vpck(vlpv,1); [typ,dum2,dum3,npts] = minfo(vlpv);end[typ,xrow,xcol,numx] = minfo(xmat);if typ == 'cons' xmat = vpck(xmat,1); [typ,xrow,xcol,numx] = minfo(xmat);end[typ,yrow,ycol,numy] = minfo(ymat);if typ == 'cons' ymat = vpck(ymat,1); [typ,yrow,ycol,numy] = minfo(ymat);endif (xcol ~= yrow) | (xcol ~= ycol) | (xrow > xcol) error('Inconsistent dimensions in X and/or Y');endviv = getiv(vlpv);% Analyze vnu,fgparm,gradfg dimensions[dum1,nbasis,ncols,npts2] = minfo(fgparm);if ncols ~= 1 error('Basis functions must be vectors.');end[dum1,ngbasis,nparmg,npts3] = minfo(gradfg);Nq = npts2/npts;Npts = int2str(Nq*npts)nparmg = nparmg/2;[type,nparm,nbnds,npts1] = minfo(vnu);if type ~= 'vary' vnu = vpck(kron(ones(npts*Nq,1),vnu),kron(getiv(fgparm))); [type,nparm,nbnds,npts1] = minfo(vnu);endif nbnds > 2 GRID_PARMV = 1;else GRID_PARMV = 0; if nbnds == 1, vnu = sbs(mscl(vnu,-1),vnu); endendnparm = nparm/2;% Check for dimension inconsistenciesif all(ngam ~= [1 Nq npts*Nq]) error('Number of gammas inconsistent with grid data')elseif any([nbasis ngbasis] ~= NX+NY) | (numx ~= NX) | (numy ~= NY) error('Number of basis functions inconsistent.');elseif any(nparmg ~= [nparm length(psi)]) error('Number of parameters inconsistent.');elseif any(npts1 ~= [npts2 npts3]) | (round(Nq) ~= Nq) disp(['There are ' int2str(npts) ' points in vlpv']) disp(['There are ' int2str(npts1) ' points in vnu']) disp(['There are ' int2str(npts2) ' points in fgparm']) disp(['There are ' int2str(npts3) ' points in gradfg']) error('Number of grid points inconsistent.');else fparm = sel(fgparm,1:NX,':'); gparm = sel(fgparm,NX+(1:NY),':'); gradf = sel(gradfg,1:NX,':'); gradg = sel(gradfg,NX+(1:NY),':');end% Identify parameters having finite, nonzero rate boundsparmx = any(vunpck(gradf)) & any(vunpck(vtp(vnu)));parmx = parmx(1:nparm) | parmx(nparm+(1:nparm));parmy = any(vunpck(gradg)) & any(vunpck(vtp(vnu)));parmy = parmy(1:nparm) | parmy(nparm+(1:nparm));nparmx = sum(parmx);nparmy = sum(parmy);if GRID_PARMV & nparmx > 0 nvertx = nbnds;else nvertx = 2^nparmx;endif GRID_PARMV & nparmy > 0 nverty = nbnds;else nverty = 2^nparmy;end% Get matrix containing all combinations of +/- for nparm-dim vector.combmatx = corners(nparmx); % combmatx = 1 if nparmx = 0combmaty = corners(nparmy); % combmaty = 1 if nparmy = 0% Get more problem dimensions[type,no,ni,nx] = minfo(xtracti(vlpv,1,1));nx1 = nx-nsf; nx2 = nsf;ny1 = nmeas-nsf; ny2 = nsf;nd = ni-nctrl; nd1 = nd-ny1;ne = no-nmeas; ne1 = ne-nctrl;eigx = zeros(npts*Nq,nvertx);eigy = zeros(npts*Nq,nverty);eigxy = zeros(npts*Nq,1);fail = [];for i = 1:npts% Get state-space data for parm_i. sys = xtracti(vlpv,i,1); [A,Bd,Bu,Ce,Cy,Ded] = transfr(sys,nmeas,nctrl,nx2); [trow,tcol]=size(Ded); if min(eig(eye(tcol)-Ded'*Ded)) <= 0 disp('I - Ded*Ded < 0'); end B11 = Bd(1:nx1,1:nd1); B12 = Bd(1:nx1,nd1+1:nd); B21 = Bd(nx1+1:nx,1:nd1); B22 = Bd(nx1+1:nx,nd1+1:nd); Ce1 = Ce(1:ne1,:); Ce2 = Ce(ne1+1:ne,:); Cy1 = Cy(1:ny1,1:nx1); Ded11 = Ded(1:ne1,1:nd1); Ded12 = Ded(1:ne1,nd1+1:nd); Ded21 = Ded(ne1+1:ne,1:nd1); Ded22 = Ded(ne1+1:ne,nd1+1:nd); Ahat = A-Bu*Ce2; Bdhat = Bd-Bu*[Ded21 Ded22]; if ny1 == 0 % (state feedback only) Atld11 = A(1:nx1,1:nx1); Atld21 = A(nx1+1:nx,1:nx1); Cetld = Ce(:,1:nx1); else Atld11 = A(1:nx1,1:nx1) - B12*Cy1; Atld21 = A(nx1+1:nx,1:nx1) - B22*Cy1; Cetld = Ce(:,1:nx1) - [Ded12;Ded22]*Cy1; end % Cycle through q gridfor ii = 1:Nq iii = Nq*(i-1)+ii; if (ngam == Nq) igam = ii; elseif (ngam == npts*Nq) igam = iii; else igam = 1; end gamma = gam(igam);% Form the Lyapunov matrices at parm_i nu = xtracti(vnu,iii,1); fdat = xtracti(fparm,iii,1); gdat = xtracti(gparm,iii,1); gfdat = xtracti(gradf,iii,1); ggdat = xtracti(gradg,iii,1); X = zeros(nx1,nx); Y = zeros(nx,nx); for k = 1:NX X = X + fdat(k) * xtracti(xmat,k,1); end for k = 1:NY Y = Y + gdat(k) * xtracti(ymat,k,1); end X11 = X(:,1:nx1); X12 = X(:,nx1+1:nx); % Check state feedback LMIs for j = 1:nverty ly = 0; parmv = zeros(2*nparm,1); for l = 1:nparm ll = nparm + l; if parmy(l) ~= 0 ly = ly + 1; if GRID_PARMV parmv(l) = nu(l,j); parmv(ll) = nu(ll,j) * psi(l); else parmv(l) = (nu(l,1)+nu(l,2))/2 + combmaty(j,ly)*(nu(l,1)-nu(l,2))/2; parmv(ll) = (nu(ll,1)+nu(ll,2))/2 + combmaty(j,ly)*(nu(ll,1)-nu(ll,2))/2; end end end Ydot = zeros(nx,nx); for k = 1:NY Ydot = Ydot + (ggdat(k,:) * parmv) * xtracti(ymat,k,1); end LMI_Y = [Y*Ahat' + Ahat*Y - Ydot - gamma*Bu*Bu', Y*Ce1', Bdhat; Ce1*Y, -gamma*eye(ne1), [Ded11 Ded12]; Bdhat', [Ded11 Ded12]', -gamma*eye(nd)]; eigy(iii,j) = max(eig(LMI_Y)); if eigy(iii,j) >= 0 disp(['Grid point ' int2str(iii) ' of ' Npts ' failed']); fail = [fail;[iii j 2]]; end end % Check output injection LMIs for j = 1:nvertx lx = 0; parmv = zeros(2*nparm,1); for l = 1:nparm ll = nparm + l; if parmx(l) ~= 0 lx = lx + 1; if GRID_PARMV parmv(l) = nu(l,j); parmv(ll) = nu(ll,j) * psi(l); else parmv(l) = (nu(l,1)+nu(l,2))/2 + combmatx(j,lx)*(nu(l,1)-nu(l,2))/2; parmv(ll) = (nu(ll,1)+nu(ll,2))/2 + combmaty(j,ly)*(nu(ll,1)-nu(ll,2))/2; end end end X11dot = zeros(nx1,nx1); for k = 1:NX X11dot = X11dot + (gfdat(k,:) * parmv) * sel(xtracti(xmat,k,1),':',1:nx1); end if ny1 == 0 LMI_X11 = [Atld11;Atld21]'*X' + X*[Atld11;Atld21] + X11dot; else LMI_X11 = [Atld11;Atld21]'*X' + X*[Atld11;Atld21] + X11dot - gamma*Cy1'*Cy1; end LMI_X = [LMI_X11, X*[B11;B21], Cetld'; [B11;B21]'*X', -gamma*eye(nd1), [Ded11;Ded21]'; Cetld, [Ded11;Ded21], -gamma*eye(ne)]; eigx(iii,j) = max(eig(LMI_X)); if eigx(iii,j) >= 0 disp(['Grid point ' int2str(iii) ' of ' Npts ' failed']); fail = [fail;[iii j 1]]; end end % Check spectral radius LMI LMI_XY = [Y, [eye(nx1);zeros(nx2,nx1)]; [eye(nx1) zeros(nx1,nx2)], X11]; eigxy(iii) = min(eig(LMI_XY)); if eigxy(iii) <= 0 disp(['Grid point ' int2str(iii) ' of ' Npts ' failed']); fail = [fail;[iii 1 0]]; endend % q loopend % p loop
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -