曝光臺 注意防騙 網曝天貓店富美金盛家居專營店坑蒙拐騙欺詐消費者

6.7. CONCLUSION
113
Table
6.5
contains
the
d.square sum error across ten test signals for each of the sigmoid methods, as well as the linear and the non-parametric method (NP). From this, it appears as if the non-parametric method is the most interesting. Evolutionary optimization and Trust Region using pre-de.ned order have slightly lower square sum error, but has the clear disadvantage of not optimizing function order. This makes the methods less reliable for signals with variable length and complexity. The non-parametric method is thus the favored progression analysis tool, as it produces repeated good results, and has superior computational speed.
Set  EO  TR  RSV  IEO  TRD  Line  NP 
1  41.853  26.6643  33.4888  54.2973  37.3468  25.5291  23.6302 
2  7.696  9.1261  9.2098  55.9464  9.3499  50.1149  10.2 
3  7.9776  7.4386  54.6359  33.9158  56.9523  14.8669  13.3758 
4  20.9424  20.1561  19.1953  73.9504  22.7163  124.2189  42.2564 
5  7.641  14.9461  71.0944  94.8918  68.3208  72  11.7348 
6  20.0249  23.8039  84.5186  64.5285  9.9265  67.8417  16.1921 
7  27.3242  30.1642  57.9581  2550.4486  50.9997  60.3503  18.0125 
8  19.3812  4.7417  4.7386  76.1091  36.4489  62.1104  17.3813 
9  2.3191  9.2118  9.2118  18.9621  8.0787  34.2552  11.0039 
10  20.6041  15.2271  8.5354  13.0756  5.1164  30.4918  9.2168 
μ  17.5763  16.1480  35.2587  303.6126  30.5256  54.1779  17.3004 
σ  11.1148  8.2909  28.0450  749.3444  21.7045  29.7517  9.3080 

Table 6.5: Progression analysis method comparison chart.
Acronyms: Evolutionary Optimization (EO), Trust Region (TR), Resid-ual Spectrum Validation (RSV), Iterative Evolutionary Optimization (IEO), Trust Region using Band-Limited Di.erentiator Pre-Processing (TRD), Line model (Line), Non-Parametric Progression Analysis (NP).
Chapter 7

 

Fault Detection
7.1 Introduction
This chapter will focus on fault detection based on trend-based feature ex-traction methods. Anomaly detection methods are developed both for the parametric and the non-parametric features. The objective of these detection methods is to identify abnormal indicator behavior without a priori knowl-edge of speci.c fault signatures. Although a framework for fault recognition is suggested, diagnosis is given lower priority. This because there is not suf-.cient training data to cover all failure modes, thus making training and validation of such a classi.cation system di.cult. Further, it is from an op-erational point of view su.cient to perform a go / no-go decision and a crude fault localization. Should a component be suspected faulty, the aircraft will in any case be subject to a through manual inspection.
　
中國航空網 www.k6050.com
航空翻譯 www.aviation.cn
本文鏈接地址：OPTIMIZATION OF FAULT DIAGNOSIS IN HELICOPTER HEALTH AND USA(62)