On the identification Volterra kernels in integral models of linear nonstationary dynamical systems

In this paper, we propose an identification algorithm for a nonstationary linear dynamical system. Conceptually, this algorithm is based on the use of piecewise linear test signals and the reduction of the original problem to a Volterra integral equation of the first kind with two variable integration limits. The numerical implementation of this algorithm is based on the quadrature formula of the middle rectangles and the product integration method. The convergence of the method of middle rectangles for a new class of linear Volterra integral equations is examined.