PROBABILISTIC AIRCRAFT CONFLICT DETECTION Eulalia Hernndez Romero - - PowerPoint PPT Presentation

PROBABILISTIC AIRCRAFT CONFLICT DETECTION IN THE TERMINAL MANEUVERING AREA

Eulalia Hernández Romero 2017 Halaby Fellow

TABLE OF CONTENTS

• I. Introduction

Slides 3-4

• II. Problem formulation

Slides 5-7

• III. Results

Slides 8-12

• IV. Summary

Slide 13

INTRODUCTION

• 1. MOTIVATION

▪ The study of the uncertainty present in the Air Traffic Management (ATM) system is a key factor to improve its levels of capacity, efficiency and safety. ▪ Sources of uncertainty: ▪ Uncertainty in data and sensors ▪ Decisions taken by individuals ▪ Weather uncertainty ▪ This project tackles the effects of wind uncertainty on the problem of aircraft conflict detection (CD) in the Terminal Maneuvering Area (TMA)

• f an airport.

INTRODUCTION

• 2. APPROACH

▪ How to study the effects on wind uncertainty on the CD problem?

• 1. Model the aircraft trajectories as random variables
• 2. Propagate weather uncertainty into the trajectory prediction

▪ Aircraft conflict indicators: ▪ Distance of closest approach ▪ Probability of conflict ▪ Conflict starting time ▪ Conflict duration ▪ … ▪ The wind uncertainty will translate into uncertain conflict indicators

PROBLEM FORMULATION

• 1. HYPOTHESES

▪ A North-East reference system is used ▪ Two aircraft, A and B, fly with approaching 3D trajectories in the same airspace ▪ The initial positions are certain ▪ An aircraft’s course, airspeed and vertical speeds is known and constant. ▪ The aircraft are affected by the same uncertain horizontal wind. The vertical component of the wind is considered negligible. ▪ The wind is defined by its two components (𝑥𝑦 and 𝑥𝑧) and it is dependent on the altitude.

PROBLEM FORMULATION

• 2. CONFLICT DETECTION

𝑄

𝐵 =

Ԧ 𝑦| Ԧ 𝑡𝐵 − Ԧ 𝑦 < 1 2 (𝑦, 𝑧, 𝑨) = 𝑛𝑏𝑦 𝑦2 + 𝑧2 𝐸 , 𝑨 𝐼 ▪ The distance between A and B is ∆(𝐵, 𝐶) = Ԧ 𝑡𝐶 − Ԧ 𝑡𝐵 ▪ There is a conflict between aircraft A and B if 𝑛𝑗𝑜 ∆(𝐵, 𝐶) = 𝜀 𝐵, 𝐶 < 1 ▪ A loss of separation is defined when an aircraft violates the protected zone of another aircraft 1000 ft 5 NM

PROBLEM FORMULATION

• 3. PROBABILISTIC ANALYSIS

Probabilistic wind model

▪ The wind components are defined as random processes:

𝑥𝑦 𝑨 = ഥ 𝑥𝑦 𝑨 + 𝜀𝑥𝑦 𝑨 𝑦 𝑥𝑧 𝑨 = ഥ 𝑥𝑧 𝑨 + 𝜀𝑥𝑧 𝑨 𝑧 ▪ The random variables x and y range from -1 to 1 𝑔

𝑦 𝑦 ,

𝑦 ∈ −1,1 𝑔

𝑧 𝑧 ,

𝑧 ∈ −1,1

Uncertainty propagation method Probabilistic Transformation Method 𝑔

𝑦

𝑔

𝑧

𝑔

𝜀 𝐵,𝐶

𝑄

𝑑𝑝𝑜

RESULTS

• 1. CONFLICT SCENARIO: DENVER AIRPORT 34L APPROACH

▪ Descending aircraft approaching a common navigation point. ▪ Simplified 1 segment trajectory

RESULTS

• 2. WIND DATA: HIGH RESOLUTION RAPID REFRESH (HRRR)

▪ Wind forecast for Nov 13th 2017 12:00 UTC forecast lead time 2h at 11000ft, merging point AJAKS (39º25’37’’N, 104º42’5’’W)

RESULTS

• 3. PROBABILISTIC WIND MODEL

▪ Vertical wind profile: mean and dispersion. ▪ Local Eulerian Probabilistic Forecast

RESULTS

• 4. CONFLICT DETECTION: DETERMINISTIC ANALYSIS

▪ Aircraft distance at closest approach: 𝑒 = 5.02 𝑂𝑁, ℎ = 1001 𝑔𝑢 ▪ No conflict: 𝜀 𝐵, 𝐶 > 1

RESULTS

• 5. CONFLICT DETECTION: PROBABILISTIC ANALYSIS

Aircraft distance PDFs ▪ Probabilistic analysis: 𝑙 𝑄

𝑑𝑝𝑜

20 km 33.4% 30 km 37.4% 40 km 40.4% The probability of conflict depends on the wind uncertainty

↑ 𝑄

𝑑𝑝𝑜

↑ 𝜀𝑥

𝑄

𝑑𝑝𝑜

SUMMARY

CONCLUSIONS

▪ Studied propagation of wind uncertainty to the problem of aircraft conflict detection in the TMA ▪ The Probabilistic Transformation Method has been successfully applied. ▪ The probability of conflict has been computed for a conflict scenario and altitude dependent winds, for different wind uncertainties.

FURTHER WORK

▪ Apply the analysis to segmented trajectories ▪ More realistic wind profiles: ▪ Time and horizontally dependent ▪ Different probabilistic characterization ▪ Develop a probabilistic resolution strategy to the TMA with the

• bjective of decreasing the

probability of conflict.

THANKS FOR YOUR ATTENTION