MATLAB Answers

S-function builder output: type mismatch error

2 views (last 30 days)
Hi everyone,
I am creating a discrete S-function that has 4 inputs, 12 outputs and 12 discrete states, with sample time of 0.02s.
Each output and input has a name associated to it and therefore I can use those name directly in my computation. The Output depends on a rotation matrix which takes value from the states in the s-function. In the mdlOutput section, the outputs have 'real_T *' type but the previous state has 'real_T' type. Therefore there is a clash of type.
In the mdlUpdate section, the results from the outputs would be stored as the next state. Again, there is a data type clash. The codes are as follows:
mdlOutput
// Angle
real_T phi = xD[3];
real_T the = xD[4];
real_T psi = xD[5];
// Rotation matrix
real_T R[3][3] = {{cos(the)*cos(psi), sin(phi)*sin(the)*cos(psi)-cos(phi)*sin(psi), cos(phi)*sin(the)*cos(psi)+sin(phi)*sin(psi)}, ...;
real_T Rt[3][3] = {{cos(the)*cos(psi), cos(the)*sin(psi), -sin(the)}, ...;
// Inertia
real_T Ixx = 0.0258;
real_T Iyy = 0.0258;
real_T Izz = 0.0502;
// Gravity
real_T g = 9.81;
// Quadrotor
real_T d = 0.29; // Distance of rotor hub from CoG in x/y direction (m)
real_T rho = 1.225; // Density of air (kg/m^3)
real_T M = 1; // Mass of quadrotor (kg)
real_T r = 0.127; // Radius of prop (m)
real_T A = 3.14159*pow(r,2); // Area of prop spinning (m^2)
real_T Ct = 0.001269; // Thrust coefficient
real_T Cq = Ct * sqrt(Ct/2); // Drag coefficient
real_T b = Ct * rho * A * pow(r,2);
real_T k = -Cq*rho*A*pow(r,3);
// Force matrix
real_T dyn[4][4] = {{-b/M, -b/M, -b/M, -b/M}, ...;
// States
z1 = R[0][0]*xD[7] + R[0][1]*xD[8] + R[0][2]*xD[9];
z2 = R[1][0]*xD[7] + R[1][1]*xD[8] + R[1][2]*xD[9];
z3 = R[2][0]*xD[7] + R[2][1]*xD[8] + R[2][2]*xD[9];
n1 = xD[9];
n2 = xD[10];
n3 = xD[11];
v1 = R[0][2]*g;
v2 = R[1][2]*g;
v3 = R[2][2]*g + dyn[0][0]*pow(u[0],2) + dyn[0][1]*u[1] + dyn[0][2]*u[2] + dyn[0][3]*u[3];
o1 = 1/Ixx * dyn[1][0]*pow(u[0],2) + dyn[1][1]*u[1] + dyn[1][2]*u[2] + dyn[1][3]*u[3];
o2 = 1/Iyy * dyn[2][0]*pow(u[0],2) + dyn[2][1]*u[1] + dyn[2][2]*u[2] + dyn[2][3]*u[3];
o3 = 1/Izz * dyn[3][0]*pow(u[0],2) + dyn[3][1]*u[1] + dyn[3][2]*u[2] + dyn[3][3]*u[3];
mdlUpdate:
xD[0] = z1;
xD[1] = z2;
xD[2] = z3;
xD[3] = n1;
xD[4] = n2;
xD[5] = n3;
xD[6] = v1;
xD[7] = v2;
xD[8] = v3;
xD[9] = o1;
xD[10] = o2;
xD[11] = o3;
Any help on how to overcome the datatype clashes is greatly appreciated.

  0 Comments

Sign in to comment.

Accepted Answer

Wern-Juin Choy
Wern-Juin Choy on 5 Aug 2014
Update: I have solved the problem. It is pretty simple actually. The inputs were reduced to a 4x1 vector. The output is reduced to a 12x1 vector output. Using xD[i] = X[i], the mdlUpdate works and w[i] on the equations in mdlOutput.

  0 Comments

Sign in to comment.

More Answers (0)

Sign in to answer this question.