當今風洞測試及計算流體力學無法模擬與分析飛機遭遇亂流時氣動力、穩定性與飛航操作品質變化。此外；應用試飛數據估算氣動力導數乃基於平衡點(Equilibrium state)下之微擾理論(Small disturbance theory)；上述分析方法不適用飛機於非穩態之亂流飛行環境。本論文遂發展新的試飛資料分析計術—模糊C均值(Fuzzy C means, FCM)混合學習徑向基底類神經網路(Radial basis neural network, RBFNN)進行氣動力建模。其飛航資料來源將不受限，本論文將以某民航機於穿音速巡航遭遇亂流時，該機飛航資料紀錄器(Flight data recorder, FDR)紀錄之飛航資料。透過RBFNN類神經網路建立該機於非穩態下氣動力，並分析遭遇亂流期間流場之物理現象及瞬間穩定性與飛航操作品質。
本研究採用擴展式卡爾曼濾波器透過動態匹配，提高飛航資料品質並估算未紀錄飛航參數，如三維風場、側滑角…等飛航參數。至於估算氣動力係數所需之推力，則透過FDR紀錄之飛航資料與本研究推導發動機推力模型進行估算。繼而將飛航資料及氣動力係數，透過模糊C均值混合學習徑向基底類神經網路，建立非穩態及非線性氣動力特徵，並探討該機遭遇強烈亂流期間，該機瞬間氣動力性能衰減、靜態及動態穩定性與飛航操控品質分析。該機飛行過程中遭遇之亂流強度，將由飛航資料進行估算。
分析結果顯示，該機遭遇強烈亂流期間阻力係數急劇增加、軸向力斜率(Axial force slop)正負號改變以及該機負載因子(Load factor)超出飛航操作手冊(Aircraft fly manual, AFM)中抖振發生(Buffeting onset)臨界值；此外控制翼面控制效能降低，尤其升降舵(Elevator)及方向舵(Rudder)。綜上所述，本研究認為該機於亂流期間，可能發生流體分離與非穩態氣動力特徵，以致降低該機縱向及橫向運動阻尼與非阻尼自然頻率，並造成縱向運動中長、短週期及橫向運動中荷蘭滾瞬間動態不穩定；該機縱向飛航操作品質亦曾瞬間衰減至Level 3，於橫向飛航操作品質瞬間降低至Level 2。最後，關於未來研究方向亦予以說明。

本研究採用擴展式卡爾曼濾波器透過動態匹配，提高飛航資料品質並估算未紀錄飛航參數，如三維風場、側滑角…等飛航參數。至於估算氣動力係數所需之推力，則透過FDR紀錄之飛航資料與本研究推導發動機推力模型進行估算。繼而將飛航資料及氣動力係數，透過模糊C均值混合學習徑向基底類神經網路，建立非穩態及非線性氣動力特徵，並探討該機遭遇強烈亂流期間，該機瞬間氣動力性能衰減、靜態及動態穩定性與飛航操控品質分析。該機飛行過程中遭遇之亂流強度，將由飛航資料進行估算。
分析結果顯示，該機遭遇強烈亂流期間阻力係數急劇增加、軸向力斜率(axial force slop)正負號改變以及該機負載因子(Load factor)超出飛航操作手冊(Aircraft Fly Manual, AFM)中抖振發生(Buffeting onset)臨界值；此外控制翼面控制效能降低，尤其升降舵(Elevator)及方向舵(Rudder)。綜上所述，本研究認為該機於亂流期間，可能發生流體分離與非穩態氣動力特徵，以致降低該機縱向及橫向運動阻尼與非阻尼自然頻率，並造成縱向運動中長、短週期及橫向運動中荷蘭滾瞬間動態不穩定；該機縱向飛航操作品質亦曾瞬間衰減至Level 3，於橫向飛航操作品質瞬間降低至Level 2。最後，關於本研究之驗證與未來研究方向亦予以說明。

It is not possible with the current technology to simulate the atmospheric turbulence effects in the wind tunnel on the aerodynamics, stability and flying quality of aircraft. In addition, the aerodynamic derivatives estimation from flight test data is always based on the concept of a small disturbance theory around an equilibrium state. The above methodologies are not suitable for aircraft flight in unsteady flight conditions, such as in atmospheric turbulence. To remedy the deficiency in the existing methodologies of flight data analysis, a new method based on the Radial Basis Function Neural Network, RBFNN, with hybrid learning by Fuzzy C Means (FCM) is developed in this dissertation. Although this new method is applicable to any flight test data, to validate it in this dissertation it is applied to flight data from Flight Data Recorder, FDR, of civil transport aircraft to model the unsteady aerodynamics. The numerical results are used to verify the physical phenomena. RBFNN describes the unsteady transonic aerodynamics by relating the flight variables to aerodynamic coefficients from flight data. The aerodynamic characteristics, stability and flying quality are then evaluated after the aerodynamic models are identified.
In the new method, the Extended Kalman Filter, EKF, is used to enhance recorded flight data’s quality through kinematics compatibility check and, simultaneously, estimate relevant unrecorded parameters, such as three dimensional wind, sideslip angle…etc. To evaluate the coefficients of aerodynamic forces and moments from flight dynamic equations, the required thrust force of turbofan engines is estimated with a newly established thrust model using the FDR data. Then RBFNN with FCM hybrid learning is applied to model unsteady aerodynamic characteristics from flight data and aerodynamic coefficients. The aerodynamics, static and dynamic stability, and flying quality are evaluated. The turbulence strength is also estimated from the recorded flight data.
In the region of severe turbulence, the aerodynamic phenomena include the rapidly increasing drag coefficient, the sign change of the axial force and the load factor exceeding the buffeting onset boundary in the Aircraft Fly Manual. The effectiveness of control power is found to be reduced as well, especially for the elevator and rudder. All of these phenomena can be explained by the existence of flow separation on the wing and the unsteady aerodynamic effects in general. Furthermore, the qualities of longitudinal motion and Dutch roll mode in the lateral-directional motion vary with time in encountering severe atmospheric turbulence and hence are nonlinear with the disturbance amplitude. The motions are also unstable at some time instants. The numerical results indicate that the worst flying quality for longitudinal motion decreases to Level 3, and the lateral-directional one decreases to Level 2. To further validate and improve the methodology, additional research is recommended.

[1] NTSB,’U.S. Air Carrier Operations Calendar Year 2005-Annual Review of Accident Data’, National Transportation Safety Board, NTSB/ARC-09/01, March 3, 2009.
[2]行政院飛航安全委員會(Aviation Safety Council, Taiwan)網站, http://www.asc.gov.tw/asc_en/accident_list_1.asp
[3] JSIT, ’Turbulence Joint Safety Implementation Team-Results and Analysis v.03’, Joint Safety Implementation Team, July, 2004.
[4] AFS-200,’Advisory Circular- Preventing Injuries Caused by Turbulence’, Federal Aviation Administration, AC-120-88A, January 19, 2006.
[5] Abernethy, J., Sharman, R., “Support Vector Machines for Regional Clear-Air Turbulence Prediction”, Flight Conference on Artificial Intelligence Applications to Environmental Science, 1.4, San Antonio, Jan. 14-18, 2007.
[6] FAA, “Turbulence Reporting Procedures,” 24th Meeting of the Informal Pacific ATC Co-ordinating Group(IPACG/24), Honolulu, USA, 2006.
[7] Vladislav, K., Eugene A. M., “Aircraft System Identification-Theory and Practice,” AIAA Education Serious, 2006.
[8] Aviation Safety Council, “行政院飛航安全委員會航空器飛航事故調查報告-中華航空公司CI 150D班機A300-600R型機由台北飛名古屋於日本硫黃島上空遭遇強烈晴空亂流”，ASC-AOR-06-09-001, September, 2006, p.p. 17.
[9] Buck, B. K., “Aircraft Acceleration Prediction Due to Atmospheric Disturbances with Flight Data Validation”, Journal of Aircraft, Vol. 43, No.1, January-February 2006, p.p. 72-81.
[10] Pollak, J., Lan, C. E., “Calculated Effect of Freestream Turbulence on Aerodynamic Characteristics of a Delta Wing”, AIAA Journal, Vol. 33, No. 9, September 1995, p.p. 1734-1735.
[11] Wang, Z. and Lan, C. E. "Fuzzy Logic Modeling of Lateral-Directional Unsteady Aerodynamics," AIAA Paper 99-4012, 1999.
[12] Yang, M.H., Ho, C.S., Hsiao, F.B., Lan, C. E., Kung, C. C.,” Estimation of Longitudinal Aerodynamics of a Transport Aircraft Using FDR Data Through Neural Network,” 7th International Industrial Simulation Conference(ISC 2009), June1~3, 2009, p.p.280~286.
[13] Wells, A., “Comparision of EDR, Calculated from the Gads Data Set, with Mean Flow Parameters From The Met Office Global Numerical Weather Prediction Model”, ICAO Meteorological Information Data Link Study Group, 2, January, 2005.
[14] Klein, V., Noderer, K. D., “Modeling of Aircraft Unsteady Aerodynamic Characteristics, Part2-Parameters Estimation From Wind Tunnel Data” NASA Technical Memorandum, NASA-TM-110161, April 1995.
[15] Morelli, E. A., Klevin V., “Application of System Identification to Aircraft at NASA Langley Research Center”, Journal of Aircraft, Vol.42, No.1, January-February 2005.
[16] Hui, K., Auriti, L., Ricciardi, J., “Advances in Real-Time Aerodynamic Model Identification”, Journal of Aircraft, Vol. 4, No.1, January-February 2005.
[17] 楊明浩,何慶雄,蕭飛賓,官文霖,楊宏智,”應用擴展示卡爾曼濾波器於無慣性裝置之飛航資料以識別風場危害”, 中國機械工程學會第二十五屆全國學術研討會,彰化大葉大學, 11月21,22日,中華民國97年.
[18] Iliff, K. W., ’Aircraft Parameter Estimation’, NASA Ames Research Center, Techniqual Report, H-1394, January 1987.
[19] Iliff, K. W., Lawrence T., ’Determination of Stability Derivatives from Flight Data Using a Newton-Raphson Minimization Technique’, NASA Ames Research Center, Techniqual Report, H-626, January 1972.
[20] Peyada, N. K., Ghosh, A. K., “Aircraft Parameter Estimation using Neural Network based Algorithm”, AIAA Atmospheric Flight Mechnics Conference, AIAA-2009-5941, Illinois, 10-13 Aug. 2009.
[21] Lan C. E., Brandon J. M., “Development of Nonlinear Aerodynamic Models from Flight Data and Evaluation of Tabulated Aerodynamic Models ”, AIAA Atmospheric Flight Mechanics Conference and Exhibit, AIAA-2003-5696, , Texas, Aug. 11-4, 2003.
[22] Philips, W. H., “Appreciation and Prediction of Flying Qualities”, NACA-Rep-927.
[23] Hodgkinson, J., “History of Low-Order Equivalent Systems for Aircraft Flying Qualities”, Journal of Guidance, Control, And Dynamics, Vol. 28, No. 4, July-August, 2005.
[24] DiFranco, D. A., “In-Flight Investigation of the Effects of Higher-Order Control System Dynamics on Longitudinal Handling Qualities”, Air Force Flight Dynamics Lab., AFFDL-TR-67-19, O.H., Aug. 1968.
[25] Morelli, E. A., “Real-time Parameter Estimation in the Frequency Domain”, AIAA Journal of Guidance, Control, and Dynamics, Vol. 23, No. 5, p.p 812-818, September-October, 2000.
[26] Morelli, E. A., “Low Order Equivalent System Identification for the Tu-144LL Supersonic Transport Aircraft”, AIAA Atmospheric Flight Mechanics Conference, AIAA-2000-3902, Denver, Aug. 14-17, 2000.
[27] Bihrle, W. J., Billy, B., “Design Charts and Boundaries for Identifying Departure Resistance Fighter Configurations”, NADC-76154-30, July 1978.
[28] Bihrle, W. J., Billy, B., “Departure Susceptibility and Uncoordinated Roll-Reversal Boundaries for Fighter Configurations”, Journal of Aircraft, Vol.19, No. 11, November, 1982, p.p. 897-903.
[29] Proctor, F. H., Hamilton, D. W., “Neumerical Study of a Convective Turbulence Encounter, ” 40th Aerospace Sciences Meeting & Exhibit, Virginia, USA, January 14-17, 2002.
[30] John, D., “Regarding the Accident of American Airlines Flight 587 at Belle Harbor, New York, November 12, 2001-Submission of the Allied Pilots Association to the National Transportation Safety Board”, NTSB, DCA02MA001, 2002.
[31] Foster, J. V., Cunningham, K., Fremaux, C. M., Shah, G. H., Stewart, E. C., Rivers, R. A., Wilborn, J. E., Gato, W., “Dynamics Modeling and Simulation of Large Transport Airplane in Upset Condition”, AIAA Navigation, Guidance, and Control Conference and Exhibit, AIAA 2005-5933, August 15-18, 2005.
[32] Boeing, “Aircraft Maintenance Manual-Indication/Recording Systems”, Chapter 31, Boeing 747-400 AMM, May, 2008.
[33] Eurocae Working Group 50, “Minimum Operational Performance Specification for Crash Protected Airborne Recorder Systems ED112,”, Paris, January 27, 2003.
[34] Minkler, G., Minkler, J., “Theory and Application of Kalman Filter”, Magellan Book Company, 1993.
[35] Gelb, A.,”Applied Optimal Estimation,” The M. I. T. Press, Cambridge, Mass., U.S.A., 1982.
[36] Cornman, L. B., Meymaris, G., and Limber, M., “An Update On the FAA Aviation Weather Research Program’s In Situ Turbulence Measurement and Reporting Ststem”, The 11th Conference on Aviation, Range, and Aerospace and the 22nd Conference on Severe Local Storms”, P4.3, October 3~8, Hyanniss, MA.
[37] Raza, H. Ioannou, P., and Youssef, H. M.,“Surface failure detecting for an F/A-18 aircraft using neural networks and fuzzy logic”, IEEE ICNN, 1994, pp. 3363-3368.
[38] Freitas, J. D., and Stevens, A. L., “Identification of vibrating structures and fault detection using neural network”, IEEE ICNN, 1996, pp. 2044-2048.
[39] Haykin, S., “Neural Networks a Comprehensive Foundation,” Prentice Hall International, Inc., 1999.
[40] 張斐章，張麗秋,”類神經網路,”東華書局,民國94年九月。
[41] 羅華強,”類神經網路-Matlab的應用,”高立圖書有限公司, 民國94年5月。
[42] Yang, M. H., Ho, C. S., Lan, C. E., Hsiao, F. B., “Longitudinal Handling Quality Analysis of a Civil Transport Aircraft Encountering Turbulence,” Journal of Aircraft, Vol. 47, No. 1, p.p. 32-40, doi: 10.2514/1.44631.
[43] Eshelby, M. E., “Aircraft Performance Theory and Practice,”Arnold , a member of the Hodder Headline Group, 2000.
[44] 楊明浩，官文霖，何慶雄，蕭飛賓，楊宏智, ”應用擴展式卡爾曼濾波器於無慣性裝置之飛航資料以識別風場危害”, 中國機械工程學會第二十五屆全國學術研討會,CSME25-870,大葉大學,彰化,2008年11月21、22日.
[45] Mattingly, J.D., Elements of Gas Turbine Propulsion, McGraw-Hill, Inc., Virginia, 1996. pp. 213-337.
[46] Roskam, J., ”Airplane Design-Part V Component Weight Estimation”, DARcorporation, 1999,p.p.17-24.
[47] Nelson, R. C., “Flight Stability and Automatic Control,” McGRAW-HILL Book Company, 1989.
[48] Brian, L. S., Frank, L. L., “Aircraft Control and Simulation,”John Wiley & Sons, Inc., 1992.
[49] Lan, C. E., Brandon, J. M., “Development of Nonlinear Aerodynamic Models From Flight Data and Evaluation of Tabulated Aerodynamic Models”, AIAA Atmospheric Flight Mechanics Conference and Exhibit, AIAA 2003-5696, August 11-14, 2003, Austin, Texas.
[50] Cook, M. V., “Flight Dynamics Principles,”Butterworth Heinemann, 1997.
[51] FAA,”Flight Test Guide For Certification of Transport Category Airlines”, AC No.:25-7A, March, 1998.
[52] Moorhouse, D. J. and Woodcock, R. J., “Background Information and User Guide for MIL-F-8785C, Military Specification-Flying Qualities of Piloted Airplane”, Flight Dynamics Laboratory, AFWAL-TR-81-3109, September, 21, 1982.
[53] ”Flying Qualities of Piloted Aircraft”, MIL-STD-1797A, Department of Defense, August 24, 2004.
[54] Hodgkinson, J.,” Aircraft Handling Quality”, Blackwell Science Ltd, Oxford, 1999, pp.54, 56.
[55] Weissman R., “Status of Design for Predicting Departure Characteristics an Spin Susceptibility”, Journal of Aircraft, Vol. 12, No. 12, December, 1975, p.p.989-993, doi: 10.2514/3.59904.
[56] Seltzer, R. M., “Investigation of Current and Proposed Aircraft Departure Susceptibility Criteria with Application to Future Fighter Aircraft”, Naval Air Development Center, NADC-90048-60, May, 1990.
[57] Ray, E. J., “Techniques For Determing Buffet Onset”, Lengley Research Center, NASA, TM X-2130, November, 1970.
[58] Friend, E. L., Sefic, W. J., “Flight Measurements of Buffet Characteristics of the F-104 Airplane For Selceted Wing-Flap Deflection”, Flight Research Center, NASA, NASA TN D-6943, August, 1972.
[59] Roskam, J., Lan, C.T., ”Airplane Aerodynamics and Performance”, DARcorporation, 1997,p.p.586-591.
[60] Larmar, W. E., “The effects of Buffeting and Other Transonic Phenomena on Maneuvering Combat Aircraft”, North Atlantic Treaty Organization, AGARD-AR-82, July, 1975.
[61] Chambers, J. R., and Hall, R. M., “Historical Review of Uncommanded Lateral-Directional Motions at Transonic Conditios”, AIAA 41st Aerospace Sciences Meeting & Exhibit, AIAA 2003-0590, January, 6-9, 2003.
[62] Chambers., J., R., Anglin, E. L., “Analysis of Lateral-Directional Stability Characteristics of a Twin-Jet Fighter Airplane at High Angles of Attack”, Langley Research Center, NASA, NASA TN-D-5361, August, 1969.