Suppose I have a dynamic system in the form: $$ a \, \dot{x}_1 + b \, \dot{x}_2 + c \,x_1 + d \, x_2 + e = u_1 \\ f \, \dot{x}_1 + g \, \dot{x}_2 + h \,x_1 + i \,x_2 + j = u_2 $$ How would I transform this system into a state space representation?

I would usually isolate for \$\dot{x}_1\$ or \$\dot{x}_2\$, but in this case, they are both a function of each other, and this is a MIMO system. Would one way be to rewrite these equations in matrix form and isolate for the \$[\dot{x}_1; \, \dot{x}_2]\$ vector (assuming matrix \$[a \; b; f \; g]\$ is not singular of course)?


You have

$$ a \, \dot{x}_1 + b \, \dot{x}_2 + c \,x_1 + d \, x_2 + e = u_1 \\ f \, \dot{x}_1 + g \, \dot{x}_2 + h \,x_1 + i \,x_2 + j = u_2 $$ That can be re written as

$$ a \, \dot{x}_1 + b \, \dot{x}_2 + c \,x_1 + d \, x_2 = u_1-e \\ f \, \dot{x}_1 + g \, \dot{x}_2 + h \,x_1 + i \,x_2 = u_2-j $$ And then $$ H = \begin{bmatrix} a & b \\f & g\end{bmatrix}, \quad -A = \begin{bmatrix} c & d \\h & i\end{bmatrix}, \quad \tilde{u} = \begin{bmatrix} u_1 -e \\ u_2 -j \end{bmatrix} \text{ and } x = \begin{bmatrix}x_1 \\ x_2\end{bmatrix}.\\ H \dot{x} - Ax = \tilde{u},\\ H \dot{x} = Ax + \tilde{u}. $$ That representation should be somewhat familiar to you except that it is a descriptor representation due to the \$H\$ matrix. If that matrix is invertible you will be able to find the usual \$\dot{x} = Ax+Bu\$. But if it is not invertible then you should dig into descriptor representation theory and find out when you can find a state-space representation to them. Or read about Differential-algebraic system of equations. I guess that \$H\$ having a left-inverse would be a must, but I don't know a thing about the descriptor representation.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.