A new method for multiparameter robust stability distribution analysis of linear analog circuits

Abstract

A correlation-first bisection method is proposed for analyzing the robust stability distribution of linear analog circuits in the multi-parameter space. This new method first transfers the complex multi-parameter robust stability problem into nonlinear inequalities by the Routh criterion, and then solves them by interval arithmetic and new bisection strategy. The axis with strong relationship to the functions dominating the stability is bisected. Furthermore, the Monte Carlo method is adopted for the uncertain subdomains to increase the convergence speed of bisection methods as the cube number increases. The proposed method has no error in both stable and unstable areas, and high efficiency to determine the complex boundaries between the stable and unstable areas. Numerical results validate this new method.