izpis_h1_title_alt

Prediction of aircraft trajectories for air traffic control using machine learning approaches
ID HRASTOVEC, MARKO (Author), ID Solina, Franc (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (4,42 MB)
MD5: 5F727CE5817204F52A67D48B8B14672E

Abstract
Air traffic is facing great challenges for the future. The economic crisis has brought a burden of cost savings, while the increase of traffic requires investments in research and development to find new paradigms for safe operations. One of the most important aspects in all future plans is better trajectory calculation, or better knowledge where the aircraft is going to be at a certain time. When positions are known, the planning can optimize flying paths to be cost efficient and safe, which is very important as the traffic becomes denser every day. Aircraft operators are planning flight paths with minimum costs, but they are not optimizing them for conflicts with other aircraft, and for airspace optimizations. Air traffic control and airspace restrictions are taking care of that. Soon, this present model will not provide enough throughput for all aircraft that want to fly. Our research is putting a stone in the mosaic of trajectory prediction and airspace optimization. In the future, aircraft will share data about their planned paths with air traffic control and aircraft in vicinity. Since air traffic is a highly regulated and expensive business, it takes a very long time before changes are implemented. Until then, we have to find alternative ways for better trajectory predictions, which will allow us to plan and optimize traffic, and to increase throughput. The ground control records the data about actual flight paths acquired by radars. Some weather data can be also acquired with a new generation of Mode-S radars. Pure aircraft performance data are enriched with weather and flight plan data into a joint knowledge database. For every new flight, we search in the database for flights similar to the incoming one. If we know how similar flights behaved in the past, we can predict the performances of a new flight, and can calculate the planned flight trajectory more accurately. Our goal is to predict trajectories better than using static models of aircraft performances. With existing prediction methods we predict for the same type of aircraft on a specific path the same trajectory every time. In that way, we have a prediction that deviates the least from the majority of flights. On the other hand, we predict a trajectory that does not fit any flight. With our approach, we want to take into account other factors such as aircraft operator, final destination, time of flight, etc., and every time predict a different trajectory suited to fit exactly to the considered flight. Operator and similar attributes are all factors that do not influence the flight directly. The destination, for instance, determines the distance of flight and therefore determines, how much fuel is on-board. More fuel means more weight and different flight characteristics. Similarly, we can assume that each operator operates airplanes differently than others, or carries different type of passengers that have usually more or less luggage than others. All these factors are not very well measurable, but they do affect flight performances. We use machine learning to find the flights in the database that are the closest to the one being predicted. With the assumption that flights with similar features flight similarly, we expect to predict more accurate trajectories than with static models and default parameters. We tested many machine learning methods and found the ones that perform the best on our data. We also adapted standard machine learning algorithms for our needs and large amounts of data. We have used machine learning predictions instead of static nominal values in widely used trajectory calculation model. The methods using only aircraft type are widely used in aviation, but they lack the capability to adapt to each flight individually. In our opinion, such rigid and static usage of aircraft type is an important cause for poor predictions. The results show that our predictions methods using individually customized predictions are more accurate than predictions based on aircraft type. We have shown that our methods are comparable with standard machine learning methods. The solution that we propose, is deployed as a web service, to which users can send flight details and get back parameters suited for a particular flight. Because the parameters are in the same form as in the widely used Base of Aircaft Data (BADA) model, legacy air control applications could use this service instead of static BADA database, and improve their trajectory calculations. In that way, a minimal change of the air control applications is needed. Trajectory calculations can remain unchanged, but with better input parameters, they can predict more accurately.

Language:English
Keywords:aircraft trajectories, BADA model, machine learning
Work type:Doctoral dissertation
Organization:FRI - Faculty of Computer and Information Science
Year:2018
PID:20.500.12556/RUL-106016 This link opens in a new window
Publication date in RUL:14.01.2019
Views:1542
Downloads:319
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:Slovenian
Title:Napovedovanje letalskih trajektorij za potrebe kontrole zračnega prometa s pristopi strojnega učenja
Abstract:
Zračni promet se spopada z velikimi izzivi za prihodnost. Zaradi ekonomske krize so se povečali pritiski na znižanje stroškov, medtem ko se na drugi strani pojavljajo zahteve za vlaganja v raziskave in razvoj, ki bodo omogočile varno delovanje ob predvidenem povečanju prometa. Eno od področij, ki se pojavlja v skoraj vseh načrtih za prihodnost, je boljše računanje trajektorij letal, oziroma boljše vedenje o tem, kje se bo letalo nahajalo ob določenem času. Ko vemo, kje letala bodo, lahko načrtujemo in optimiziramo varne zračne poti dlje v prihodnost, kar je zelo pomembno za vedno gostejši promet. Letalski prevozniki minimizirajo stroške z načrtovanjem optimalnih poti. Pri tem upoštevajo mnogo dejavnikov, kot so vremenske napovedi in omejitve v zračnem prostoru. Ne morejo pa upoštevati ostalega prometa. Kontrola zračnega prometa s pregledom nad vsemi letali skrbi za varno uporabo zračnega prostora. Kontrolorji se odločajo glede na trenutno stanje in po potrebi letala preusmerijo z načrtovane optimalne poti zaradi drugih letal v bližini. Opisan model kmalu ne bo več zmožen zagotavljati dovolj prepustnosti zračnega prostora za vsa letala, ki bi želela leteti. Naša raziskava je delček v mozaiku napovedovanja trajektorij letal in optimizacije zračnega prostora. V prihodnosti si bodo letala izmenjevala podatke o zračnih poteh s kontrolo na tleh in bodo letela po načrtu, ki bo optimiziran tudi glede na ostali promet. Ker je zračni promet zelo reguliran, vsaka sprememba vzame ogromno časa. Do predvidenih sprememb moramo uvesti majhne izboljšave v okviru trenutnega sistema, s katerimi bomo bolje napovedovali trajektorije, ki bodo omogočale načrtovanje in optimizacijo zračnega prometa ter povečanje propustnosti zračnega prostora. Dejanske poti letenja letal snemamo z radarji in jih hranimo za poznejšo uporabo. Z novimi radarji Mode-S lahko dobimo z letal tudi nekaj vremenskih podatkov. Podatke o letalskih zmogljivostih izračunamo iz shranjenih letalskih poti. Obogatimo jih še z vremenskimi podatki in načrti letov. Načrti letov vsebujejo pomembne informacije o tipu letala, prevozniku, načrtovani poti in še mnogo drugega. S tem dobimo veliko podatkovno bazo znanja o preteklih letih. Ko pričakujemo nov let, v podatkovni bazi poiščemo lete, ki so podobni temu, ki ga pričakujemo. Če znamo poiskati lete s podobnimi letalnimi karakteristikami, lahko napovemo zmogljivosti prihajajočega leta in lahko izračunamo načrtovano trajektorijo leta. S trenutno uporabljanimi metodami za izračun vedno napovemo enako trajektorijo za isti tip letala in isto pot, ker imamo na voljo le nominalne vrednosti za določen tip letala. Nominalne vrednosti so določene tako, da so najboljši približki letov, ki so jih imeli na razpolago snovalci sistema. Leti v našem zračnem prostoru pa so drugačni. Naš cilj je napovedati boljše vhodne parametre za izračun trajektorij s pomočjo vedenja o shranjenih preteklih letih in izračunati trajektorije, ki bodo bližje dejanskim potem letenja. Z našo rešitvijo upoštevamo vidike, kot so upravljavec letala, končna destinacija, čas letenja, itd., in vsakič napovemo drugačno trajektorijo, ki je prilagojena točno določenemu letu. Našteti atributi ne vplivajo neposredno na let. Končna destinacija, na primer, določa dolžino leta in s tem vpliva na to, koliko goriva bo na krovu letala. Več goriva pomeni večjo težo in drugačne letalne lastnosti. Podobno lahko sklepamo, da letalske družbe letijo različno. Nizkocenovni prevozniki običajno vozijo potnike z manj osebne prtljage, kar vpliva na težo. Vsi ti in podobni dejavniki niso lahko merljivi, a vplivajo na letalske zmogljivosti. Da bi našli lete v podatkovni bazi, ki so najbližje napovedanemu letu, uporabljamo strojno učenje. S predpostavko, da leti s podobnimi lastnostmi letijo podobno, pričakujemo, da bomo napovedovali točnejše trajektorije kot s statičnimi modeli in nominalnimi parametri. Preizkusili smo mnogo algoritmov strojnega učenja za to vrsto podatkov in našli najprimernejše. Prilagodili smo standardne algoritme strojnega učenja za naše potrebe in za veliko količino podatkov, ki jih imamo. Napovedi strojnega učenja smo namesto nominalnih vrednosti vnesli v najbolj uveljavljen model za izračun trajektorij. Metode, ki za izračune uporabljajo le tip letala, se redno uporabljajo v letalstvu, a jim primanjkuje zmožnosti, da bi se prilagodile posameznemu letu. Taka statična in toga uporaba je po našem mnenju glavni vzrok slabih napovedi. Rezultati kažejo, da so naše napovedi, ki so prilagojene posameznemu letu, natančnejše. Pokazali smo, da so rezultati naših metod primerljivi z najboljšimi standardnimi metodami strojnega učenja. Rešitev je narejena kot storitev, ki ji uporabniki lahko pošljejo podrobnosti o letu in dobijo nazaj prilagojene parametre o predvidenih zmogljivostih tega leta. Ker so parametri v enaki obliki kot v najbolj uporabljanem modelu Base of Aircraft Data (BADA), lahko obstoječe aplikacije uporabijo storitev namesto nominalnih parametrov. S tem bi izboljšale svoje napovedi le z majhnim posegom. Metode izračuna trajektorij lahko ostanejo nespremenjene. Dobile bi le boljše vhodne parametre in bi zato nudile točnejše izračune trajektorij.

Keywords:trajektorije letal, BADA model, strojno učenje

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back