曝光臺(tái) 注意防騙 網(wǎng)曝天貓店富美金盛家居專營(yíng)店坑蒙拐騙欺詐消費(fèi)者

Given
the
type
of
indicator
and
the
component
from
which
it
originates, at threshold breach gives both an indication that something is wrong, as well as information on which component is faulty and what type of failure it su.ers from. In a practical implementation, it is common to require N out of M threshold overshoots on a given indicator before an alarm is raised. This is to avoid that indicator outliers, in the context of HUMS referred to as spikes, result in unjusti.ed alarms.
The main objection to threshold testing in health monitoring is the dif-.culty in setting the optimal threshold values. Setting thresholds too low might result in false alarms, i.e. threshold overshoots despite the fact that nothing is wrong. Setting the thresholds too high renders the system less sensitive to variations in the vibration signature, and thus less equipped for detecting faults. For some indicators, it is possible to set global or .xed thresholds. This means that the same threshold is applied across an entire .eet. Unfortunately, most indicators have a normal state envelope which is unique to each aircraft. Further, this envelope is prone to change between major overhauls, a phenomenon known as a step change. To accommodate for this, thresholds must constantly be updated for each aircraft.
Threshold adjustment, or learning, is performed on new aircrafts and after major overhauls. The process consists in acquiring a statistically signi.cant baseline of observations, typically on the magnitude of 50 .ight hours, and calculating the gaussian localization μi and distribution σi parameters on the dataset. The threshold or thresholds for an indicator i are then de.ned using a threshold policy of type Ti = μi + Nσi. During the learning period, a set of alternate thresholds are used. These are global, and are to avoid false alarms set so high that they have reduced chance of detecting faults. Consequently, the aircraft is vulnerable during the training period.

Threshold re-learning is a tedious task for heavy aircraft with several hun-dred indicators, and it is not always possible to predict which overhauls will require re-learning of which indicators. This burden is a common complaint from operators who wishes more autonomous solutions.
Alternative variants are hysteresis thresholds, hypothesis testing and Bayesian decision approaches. Hysteresis thresholds are applicable in systems where it is necessary to measure the number of times a variable crosses a threshold over a given period. This method is used in several of the usage monitoring functions of the HUMS, but has no obvious applications in health monitoring.
Using hypothesis testing it is possible to compare two groups of observa-tions, and .nd the possibility of the two groups originating from the same distribution. If one group represents the normal state baseline and the other a set of observations from an asset in an unknown condition, it reasonable to assume that the asset is in a damaged state if its associated observation dis-tribution is highly di.erent from the normal state baseline. This is in reality a generalization of the threshold testing method described above, but per-mits comparing a group of samples to the learnt baseline. Another variant is analyzing the possibility of various failure modes given an alarm. By knowing these prior probabilities, it is possible to identify the most likely problem,
　
中國(guó)航空網(wǎng) m.k6050.com
航空翻譯 www.aviation.cn
本文鏈接地址：OPTIMIZATION OF FAULT DIAGNOSIS IN HELICOPTER HEALTH AND USA(24)