p(x): probability density distribution r(x): probability density ratio psuccess(x): density distribution for satisfactory samples pall(x): density distribution for all samples (x|μ,Σ): normal distribution with mean μ and covariant matrix Σ k : index of components for mixture model π k: mixing coefficient for k-th mixture component μk : mean for k-th mixture component ∑k: covariant matrix for k-th mixture component x: states of aircraft dynamics u: control input y: output w: velocity in z-direction (body axis) q: pitch angular velocity α: angle of attack αc: angle of attack command k: adjustable parameter vector Kα: feedback gain for AoA (adjustable parameter) Kq: proportional feedback gain for pitch angular velocity (adjustable parameter) KIα: integral feedback gain for AoA (adjustable parameter) Kδe: additional elevator loop gain (uncertain parameter) TD: delay time (uncertain parameter) σ: standard deviation ( ) : estimated value
[1]
Takafumi Kanamori,et al.
Density Ratio Estimation in Machine Learning
,
2012
.
[2]
Robert F. Stengel,et al.
Robust Control System Design Using Simulated Annealing
,
2000
.
[3]
Radford M. Neal.
Pattern Recognition and Machine Learning
,
2007,
Technometrics.
[4]
Yoshikazu Miyazawa,et al.
Stochastic Parameter Tuning Applied to Space Vehicle Flight Control Design
,
2001
.
[5]
Junichiro Kawaguchi,et al.
Stochastic Approach to Robust Flight Control Design Using Hierarchy-Structured Dynamic Inversion
,
2011
.