On a problem of group classification of linear dynamical systems (Q2739963)

Let us consider a linear system of automatic control \(\dot x(t)=Ax(t)+Bu(t), x(t)\in X, u(t)\in U\), where \(X,U\) are linear spaces; \(A,B\) are complex matrices; \(x(t)\) is a state of the system; \(u(t)\) is the input signal. This system is determined by matrices \((A,B)\in \mathbb{C}^{n^2+nm}=\{(A,B)\mid A\in \mathbb{C}^{n\times n}, B\in \mathbb{C}^{n\times m}\}\), \(\mathbb{C}^{n\times n}\) is the linear space of complex \(n\times n\)-dimensional matrices. A classification of \(\mathbb{C}^{n^2+nm}\) with respect to a change of the basis in \(X\) and \(U\) is proposed. The subset of pre-stable systems \(W\subset \mathbb{C}^{n^2+nm}\) is obtained. In the case \(m=2\) the factors \(\mathbb{C}^{n^2+nm}/(f)\) of the main open sets covering \(W\) are constructed. The application of the invariants theory to the problem of parametric identification of systems is presented.