<p>Input signal design is very important in academy and practice, as it affects the estimation accuracy and control performance. This paper extends our previous open loop aircraft flutter model analysis to more general closed loop case, as closed loop situation is more practical than open loop case. Due to two measured noises in this general closed loop aircraft flutter model analysis, optimal input signal is determined from the aspect of unbiased estimate, i.e. one unbiased nonparametric flutter model only through the collected close loop input–output data. To give the detailed dependence about optimal input signal on nonparametric estimate, we derive one explicit improved form and its statistical analysis. Moreover, to guarantee the equivalence between the unbiased estimate and convergence property, one composite Lyapunov analysis is formulated to consider two different noises, similar to be convergence within noise, i.e. robustness. Finally, one platform is established and some simulations are done to prove our proposed theoretical results. Generally, this paper shows our new ideas about how to improve the accurate identification within the closed loop situation and input–output noises simultaneously, more suiting to the practical aircraft flutter experiment.</p>