Robust Optimal Regional Closed-loop Pole Assignment over Positivity Conditions and Differential Evolution

摘要

The presented paper describes an optimal closed-loop pole assignment for a vast control feedback framework and structure. The main aim of this paper is a regional selection of the closed-loop central polynomial with additional flexibility and leeway for the following optimization procedure. The optimality of the approach is provided through the positivity condition of the quasi-convex spectral polynomial (QSP). The spectral polynomial presents the uncertainty and closed-loop characteristic of the feedback system. The derivation of the QSP is based on the characteristics of the uncertainty models and metrics with norm H∞. The structure of the controller is freely chosen and is also the subject of the optimization procedure. An objective function for the optimization algorithm Differential Evolution (DE) is presented in multi-criteria form and is directly composed of various QSP's. Such optimization of the objective function improves the transparence, computation expense and the regularity of the optimal/suboptimal solution.