|摘要: ||本研究將波音747型機之兩筆正常航班資料及一筆異常航班資料由飛行操作品質系統(Flight Operations Quality Assurance, FOQA)取出之飛行數據，利用小波法(Wavelet Transform)及希爾伯特-黃轉換(Hilbert-Huang Transform, HHT)加以分析。考慮許多相關程度高的參數，如空速、下降率、引擎推力、攻角、風速、風向等，從原始訊號分解出不同頻率之訊號，觀察訊號變化，研判出訊號有研究價值的部分，藉由兩個理論方法分析這些參數，比較三個航班之異常訊號。|
本研究主要利用飛行力學以及推估導航之系統概念，藉由FOQA系統裡快速資料記錄器(Flight Data Recorder, FDR)所提供的降落飛行資料，重新建立二維平面風場。再將與降落有關的參數放進兩種理論方法加以分析。首先利用HHT解析訊號，HHT是一個優秀的工具，能闡述非線性和非平穩時間序列，產生不同頻率之訊號，與傳統方法相比，在隱藏物理現象的理解上，HHT具有更高頻譜的分辨率。HHT可將原始訊號將被分解出好幾個震盪模組(Intrinsic Mode Functions, IMF)，層層抽出，更重要的是了解分離後訊號背後代表之物理意義。其次有別於以往如傅立葉轉換(Fourier transform , FT)，短時快速傅立葉轉換(Short-time Fourier transform, STFT)等固定窗下的頻率解析，小波法(Wavelet transform)套用母小波來分析一段訊號。而不同高低之頻率需求不同長短的窗型來解析，因此小波法利用Morlet型母小波調整窗大小來達到解析高低頻率之間的缺點，因此解析度將比傳統方法得到在頻域上更突出的結果。
In this research, three flight data of Boeing747-400 including two normal and one abnormal are analyzed, which derived from Flight Operations Quality Assurance (FOQA), by using Wavelet Transform and Hilbert-Huang Transform methods. Considering the tremendous amount of relevant parameters involved, such as true airspeed, vertical speed, thrust, angle of attack, wind speed, wind direction, etc., we need to first decompose these engineering data to different frequencies inside the signals, observe their changes, and acquire valuable and meaningful flight interpretations. The main purpose for this work is to figure out unusual warning by contrasting with the analyses via two theoretical methods.
In this study, we re-establish two-dimensional horizontal wind fields by data taken from Flight Data Recorder (FDR) via the equations of flight mechanics and the system of dead-reckoning. Then we could analyze parameters of wind speed and wind direction into HHT and Wavelet Transform formats. HHT is a modern tool for non-linear and non-stationary time series interpretation, and has a high-resolution spectrum compared to traditional methods for understanding background physical meaning. In the cases studied, HHT gave results much sharper than those from any of the traditional analytical methods in the time-frequency-energy representations. Moreover, traditional methods such as Fourier transform (FT) and short-time fast Fourier (STFT), utilizing fixed window to analyze frequencies in signal, but Wavelet Transform applying a dilation window function to fit the length of the data, and it really work out on decomposing from high-frequency to low-frequency oscillation in signals. Therefore, Wavelet Transform obtains more prominent results in the frequency domain.
Data generated consequences of HHT and Wavelet method, compared with each other, in order to see clearly the results which were demonstrated. Comprehensive the interpretations of two methods combined with the report of Aviation Safety Council give us more precise analytical results in landing profile and physical phenomena. Moreover, our expectation is to obtain the "prediction warning message" which would pose a serious threat to flight safety before the occurrence of accident/incident from observing the abnormal parameters data set.