Fitting experimental data to ODE add additional fitting parameter which should not be integrated by ODE-Solver

1 Ansicht (letzte 30 Tage)
Hey Matlab-Friends,
I want to fit some Experimental data to a First Order ODE Model and get the unknown parameters of this ODE. The fitting procedure (all fitting parameters are a part of the ode) is already working and I get my parameters as desired. Now I want to add another fitting parameter in my routine which is not part of my Model ODE. The fitting parameter is just a constant which should be added to each value of my ODE Solution, like a scaling factor.
Could anyone help me out with that? This would be very helpful. Here is the code:
if true
% function bestc = odeParamID()
%%Create some "experimental data"
clc; clear all; clf
filename = 'Methanol_1.txt'
data_mess = dlmread(filename,'\t',0,0);
tdata = data_mess(:,1);
mdata = data_mess(:,2)/1000;
plot(tdata, mdata)
%%ODE Information
tSpan = [0 350];
z0 = 0.000000000001; % Initial values for the state
%%Initial guess
c0 = [80,84e-9,0.74,0.00001908]; % First guess at parameter value
ODE_Sol = ode15s(@(t,z)updateStates(t,z,c0), tSpan, z0); % Run the ODE
simY = deval(ODE_Sol, tdata); % Evaluate the solution at the experimental time steps
hold on
plot(tdata, simY, '-r')
%%Set up optimization
myObjective = @(x) objFcn(x, tdata, mdata,tSpan,z0);
lb = [0,84e-10,0.69,0.00001794];
ub = [180,84e-8,0.79,0.00002024];
options = optimoptions(@lsqnonlin,'Algorithm','trust-region-reflective','OptimalityTolerance', 1.0000e-10,'FunctionTolerance',1e-10,'StepTolerance',1.0000e-10,'MaxFunEvals',10000)
bestc = lsqnonlin(myObjective, c0, lb, ub);
%%Plot best result
ODE_Sol = ode15s(@(t,z)updateStates(t,z,bestc), tSpan, z0);
bestY = deval(ODE_Sol, tdata);
plot(tdata, bestY, '-g')
legend('Exp Data','Initial Param','Best Param');
function f = updateStates(t, z, c)
g = 9.81; % m*s-2
Density = 792; %kg*m-3
SurfaceEnergy= 0.0224; %J*m-2
%c(2)r = 0.000000250; %m
%c(1)Phi= 60; %°
Viscosity= 0.000543; % kg m-2 s-1
%c(4)Surface= 0.00001;
%c(3)Porosity = 0.6;
a = 2*SurfaceEnergy/(8*Viscosity);
b = g*Density/(8*Viscosity);
q = Density;
f = a*abs(cosd(c(1)))*c(2)*q^2*c(3)^2*c(4)^2*1/z-b*c(2)^2*q*c(3)*c(4);
function cost = objFcn(x,tdata,mdata,tSpan,z0)
ODE_Sol = ode15s(@(t,z)updateStates(t,z,x), tSpan, z0);
simY = deval(ODE_Sol, tdata);
cost = simY-transpose(mdata);
end

Akzeptierte Antwort

Torsten
Torsten am 17 Nov. 2017
if true
% function bestc = odeParamID()
%%Create some "experimental data"
clc; clear all; clf
filename = 'Methanol_1.txt'
data_mess = dlmread(filename,'\t',0,0);
tdata = data_mess(:,1);
mdata = data_mess(:,2)/1000;
plot(tdata, mdata)
%%ODE Information
tSpan = [0 350];
z0 = 0.000000000001; % Initial values for the state
%%Initial guess
c0 = [80,84e-9,0.74,0.00001908,0]; % First guess at parameter value
ODE_Sol = ode15s(@(t,z)updateStates(t,z,c0), tSpan, z0); % Run the ODE
simY = deval(ODE_Sol, tdata); % Evaluate the solution at the experimental time steps
hold on
plot(tdata, simY, '-r')
%%Set up optimization
myObjective = @(x) objFcn(x, tdata, mdata,tSpan,z0);
lb = [0,84e-10,0.69,0.00001794,-inf];
ub = [180,84e-8,0.79,0.00002024,inf];
options = optimoptions(@lsqnonlin,'Algorithm','trust-region-reflective','OptimalityTolerance', 1.0000e-10,'FunctionTolerance',1e-10,'StepTolerance',1.0000e-10,'MaxFunEvals',10000)
bestc = lsqnonlin(myObjective, c0, lb, ub);
%%Plot best result
ODE_Sol = ode15s(@(t,z)updateStates(t,z,bestc), tSpan, z0);
bestY = deval(ODE_Sol, tdata);
plot(tdata, bestY, '-g')
legend('Exp Data','Initial Param','Best Param');
function f = updateStates(t, z, c)
g = 9.81; % m*s-2
Density = 792; %kg*m-3
SurfaceEnergy= 0.0224; %J*m-2
%c(2)r = 0.000000250; %m
%c(1)Phi= 60; %°
Viscosity= 0.000543; % kg m-2 s-1
%c(4)Surface= 0.00001;
%c(3)Porosity = 0.6;
a = 2*SurfaceEnergy/(8*Viscosity);
b = g*Density/(8*Viscosity);
q = Density;
f = a*abs(cosd(c(1)))*c(2)*q^2*c(3)^2*c(4)^2*1/z-b*c(2)^2*q*c(3)*c(4);
function cost = objFcn(x,tdata,mdata,tSpan,z0)
ODE_Sol = ode15s(@(t,z)updateStates(t,z,x), tSpan, z0);
simY = deval(ODE_Sol, tdata);
cost = simY+x(5)-transpose(mdata);
end
does not work ?
Best wishes
Torsten.
  3 Kommentare

Melden Sie sich an, um zu kommentieren.

Weitere Antworten (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by