For the example dataset, the approximation has one miss-placed step and a signi.cant bias toward the end of the dataset. In general, this method works well. The main drawbacks is that it does not determine order, and convergence is slow due to a poorly chosen initial condition.
Trust Region with Pre-De.ned Order (TR)
This method was tested using convergence of the solution to the approximate function as stopping criteria. The dcd parameter was given the initial value
(k)(k)
equal to μi, while the ad and qd parameters were set to 0 and 1 respectively. Each sigmoid’s position, p(dk), was set so that the sigmoids were uniformly distributed across t. Final
results
are
shown
in
.gure
6.18.
Orders was set to 3.
This method performs reasonably well on the example dataset. The main
6.4. SIGMOID PROGRESSION ANALYSIS
objection to this method is execution speed, as robust evolutionary optimiza-tion requires a large number of evaluations of the object function. Further, this method is incapable of determining model order.
Residual Spectrum Validation (RSV)
In the context of HUMS data analysis, it would be su.cient to analyze the last 200 -300 .ight hours of an aircraft to assess the condition of the drive-train. For an indicator series of this duration, more than three or four transition would be extremely unlikely. Consequently, good results can be obtained using a .xed model order. Still it would be desirable to automati-cally determine the optimal model order, with the optimal order being one which balances the power spectrum of r.while at the same time minimizes model order.
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址:OPTIMIZATION OF FAULT DIAGNOSIS IN HELICOPTER HEALTH AND USA(52)