The problem of robust D-stability analysis with respect to real convex poly
topic uncertainties is tackled. A new LMJ-based sufficient condition for th
e existence of parameter-dependent Lyapunov functions is proposed. This con
dition generalises previously published conditions. Numerical comparisons w
ith quadratic stability results as well as previous results based on parame
ter-dependent Lyapunov functions illustrate the relevance of this new condi
tion. Finally, this result appears to be promising for robust multi-objecti
ve performance analysis and control synthesis purposes. (C) 2000 Elsevier S
cience B.V. All rights reserved.