Analysis of linear systems with fuzzy parametric uncertainty (Q1874061)

The problem of determining the stability margin of a linear system with fuzzy parametric uncertainty is addressed. This kind of system arises when the exact value of a process parameter is not available, but an estimation of the value is given by an expert. As the \(\alpha\)-cut representation of a fuzzy number is a functional interval, interval analysis proves to be a suitable tool to deal with fuzziness. Based on the \(\alpha\)-cut representation of a fuzzy number and the Argoun stability test for an interval system, a set of algorithms are derived to determine how much uncertainty (maximum confidence degree) can be afforded on every coefficient of the characteristic polynomial. The main advantage of the method is that an independent confidence degree is considered for each coefficient. This leads to a clear understanding of how each coefficient affects stability, and of which coefficients have the most influence on this. Based on Argoun's stability test for interval systems, a new methodology, which has a natural graphical interpretation, is presented. An analysis on how precise some coefficients shouldbe to allow more uncertainty for others can be carried out. A comparison with the Tsypkin-Polyak plot is also presented.