Identifiability and Adaptive Identification of Decentralized Systems

The identification problem of decentralized systems (DS) is considered. The analysis shows that this issue has not been given sufficient attention. The increasing complexity of systems and a priori uncertainty require the development of approaches and methods. First, this concerns the parametric identifiability (PI) of decentralized systems. We propose the approach to the PI evaluation based on the constant excitation condition and considered relationships in the subsystems. The conditions for local structural and parametric identifiability by output and state space are obtained. Adaptive algorithms for decentralized system parametric identification have been synthesized. The exponential dissipativity of the adaptive system is proved. The exponential dissipativity of the adaptive system is proved. The obtained results are based on the application of Lyapunov vector functions. We study the influence of relationships in subsystems on the properties of parameter estimates. It is shown that an adaptive algorithm can be described as a dynamic matrix system if a functional constraint is imposed on the adaptive identification system (AIS). We consider a special case of such algorithms with a delayed argument and study the stability of an adaptive system. The influence of connections in the system on the DS identifiability is analyzed. The conditions of adaptive identification system exponential stability are obtained. We present results of mathematical modeling. The system comprising two subsystems is considered. Adaptive identification results are presented in the output space. The influence of subsystems is studied on estimates of subsystem model parameters. The structures reflecting the difference in the speed of parameter tuning of subsystem models are presented.