회전익 항공기 비행 시뮬레이터를 통한 비행착각 감지 Tool 개발에 관한 연구 = (A) study on Development of a Spatial Disorientation Detection Tool through Helicopter Flight Simulator

다국어 초록 (Multilingual Abstract)

Despite the systematic advancements in aviation industry technology and safety management, accidents of spatial disorientation in helicopter persistently occur. Spatial disorientation, a phenomenon where the pilot perceives the aircraft's attitude differently from its actual orientation, arises from multifaceted causes including fatigue, stress, and low-visibility conditions. Research dedicated to detecting flight illusions has predominantly centered on human factors like pilots' physiological attributes and behavioral responses. However, the development of a commercialized tool for detecting spatial disorientation remains unrealized. Predominantly, spatial disorientation fall under Type I category, characterized by the pilot's lack of awareness, thereby imposing constraints on the effectiveness of detecting these illusions solely based on physiological responses of pilots. Consequently, this study endeavors to develop an innovative tool for the detection and response to spatial disorientation in helicopter.
The methodology of this research involved designing scenarios that closely mirror actual flight conditions, thereby facilitating the collection of high-quality data. Subsequently, the reliability of this data was ascertained through meticulous error rate analysis. The gathered data underwent a rigorous process of feature selection criteria establishment and the formulation of data labeling algorithms, enhancing the precision of spatial disorientation classification and detection accuracy.
This study yielded significant findings, confirming that flight data encompassing control device input values, aircraft attitude values, and specification values serve as reliable standards for the precise detection of spatial disorientation scenarios. Moreover, it established that beyond the traditionally recognized factors of pitch and bank associated with spatial disorientation, parameters such as heading, pedal, collective, and aircraft velocity emerge as pivotal in detecting spatial disorientation.
Furthermore, the study employed error rate analysis and phenomenological research to substantiate that factors like a pilot's qualification level and possession of Instrument Flight Rules (IFR) qualification play a crucial role in the incidence of errors related to spatial disorientation. The research highlighted a prevalent state of psychological anxiety among helicopter pilots, attributable to complex mission environments and inadequate training. When this state is compounded with human factors such as nocturnal and moonless conditions, fatigue, and stress, the propensity for spatial disorientation escalates significantly.
The application of this tool, as proposed by the study, has the potential to enhance the accuracy in recognizing spatial disorientation conditions. Effectively responding to these conditions through improvements in pilot education and training could significantly contribute to the mitigation of helicopter accidents.

목차 (Table of Contents)

• 제1장 서론 1
• 1.1 연구의 배경 1
• 1.2 연구 필요성 및 목적 3
• 1.3 연구절차 및 방법 6
• 제2장 이론적 배경 8
• 2.1 비행착각의 이해 8
• 2.1.1 인체의 균형과 감각기관 9
• 2.1.2 비행착각의 원인 13
• 2.1.3 비행착각의 종류 15
• 2.2 비행착각 사고사례 23
• 2.2.1 국외 비행착각 사고분석 23
• 2.2.2 국내 비행착각 사고분석 26
• 2.3 비행착각 극복 훈련 28
• 2.3.1 ICAO 기준 28
• 2.3.2 FAA 기준 29
• 2.3.3 국내 기준 30
• 2.3.4 국내 항공기 운영기관 30
• 2.3.5 대한민국 공군 31
• 2.4 선행연구 32
• 2.5 용어 정의 39
• 2.5.1 비행착각 39
• 2.5.2 저시정 40
• 2.5.3 비정상 자세 41
• 2.5.4 항공기 사고 42
• 제3장 연구모형 및 방법 44
• 3.1 연구 모형 44
• 3.2 현상학적 분석 47
• 3.3 비행 데이터 신뢰성 검증 51
• 3.4 머신러닝 알고리즘 52
• 제4장 연구결과 54
• 4.1 시뮬레이터 시나리오 설계 54
• 4.1.1 주제, 주제군 및 범주 분류 54
• 4.1.2 도출된 자료에 대한 기술 58
• 4.1.3 시뮬레이터 시나리오 설계 69
• 4.1.4 현상학적 연구 결과 74
• 4.2 비행 데이터 신뢰성 검증 78
• 4.2.1 가시거리 79
• 4.2.2 비행착각 84
• 4.2.3 조종사 자격 91
• 4.2.4 비행단계 99
• 4.3 비행착각 감지 Tool 개발 106
• 4.3.1 데이터 수집 및 전처리 106
• 4.3.2 특징 선택 기준설정 108
• 4.3.3 데이터 레이블링 110
• 4.3.4 의사결정트리(Decision Tree) 모델 112
• 4.3.5 랜덤 포레스트(Random Forest) 모델 114
• 4.3.6 엑스트라 트리(Extra Trees) 모델 117
• 4.3.7 그래디언트 부스팅(Gradient Boosting) 모델 119
• 4.3.8 비행착각 감지 Tool 성능 평가 122
• 제5장 결론 및 논의 125
• 5.1 연구결과 및 기대효과 125
• 5.2 연구의 한계 및 향후 연구 방안 128
• 참고 문헌 129
• 부록 140
• SUMMARY 154