The next gen­er­a­tion of cloud solutions are here to help process data from pre­dic­tive main­te­nance to in-flight per­for­mance and ag­ing of the air­craft can be bet­ter tracked and un­der­stood

SP's Airbuz - - Table of Contents - BY SUKHCHAIN SINGH

The next gen­er­a­tion of cloud solutions are here to help process data from pre­dic­tive main­te­nance to in-flight per­for­mance and ag­ing of the air­craft can be bet­ter tracked and un­der­stood

WHEN “AR­TI­FI­CIAL IN­TEL­LI­GENCE” (AI) and avi­a­tion are thought of, we think of drones. Au­ton­o­mous air­craft are how­ever, only a frac­tion of the im­pact that ad­vances in ma­chine learn­ing and other AI tech­nolo­gies have in avi­a­tion. Air­craft man­u­fac­tur­ers and air­lines are in­vest­ing heav­ily in AI tech­nolo­gies with ap­pli­ca­tions that range from the flight­deck to customer’s ex­pe­ri­ence and MRO sphere.

AI IN MRO. Rolls-Royce has in­tro­duced the SWARM ro­bots. These minia­ture ro­bots can be in­serted within the cen­tre of an en­gine and then these crawl into hard-to-ac­cess ar­eas to carry out vis­ual in­spec­tion. The cam­era-equipped ro­bots send live video to the op­er­a­tor, who can quickly check the en­gine for prob­lems. The FLARE is another so­phis­ti­cated di­ag­nos­tic tool that con­sists of two snake-like ro­bots that can travel through­out the en­gine and carry out patch re­pairs. The fi­nal one of Rolls-Royce’s new toys is a re­mote bore blend­ing ma­chine that can be in­stalled on the en­gine and then con­trolled re­motely by one of Rolls-Royce’s own trained

tech­ni­cians to do spe­cialty work like grind­ing parts with lasers.

With minia­turised sen­sors and mo­tion con­trol de­vices now avail­able, is it time to use the ad­van­tages of dig­i­tal tools to change the way that aero­plane en­gines are main­tained? This forms the ba­sis of Rolls-Royce’s In­tel­li­gent En­gine pro­gramme, which aims to use the de­vel­op­ments in hard­ware and dig­i­tal tech­nol­ogy to fuse prod­uct and ser­vice to­gether.

‘In­spect’ is a net­work of ‘periscope’ op­ti­cal sen­sors per­ma­nently em­bed­ded within the en­gine, en­abling it to in­spect it­self to spot and re­port any main­te­nance re­quire­ments, which is re­ported back to the op­er­a­tions cen­tre. These pen­cil-sized ro­bots are pro­tected from the ex­treme heat gen­er­ated within the en­gine and the vis­ual data cre­ated is used along­side the mil­lions of data points al­ready gen­er­ated by today’s en­gines as part of their en­gine health mon­i­tor­ing sys­tems.

Lo­cal staff insert de­vices that pro­vide in­for­ma­tion, are con­trolled from a sin­gle site and re­pairs are car­ried out re­motely. Sav­ings in time and money would be con­sid­er­able, but more im­por­tantly, it would en­sure max­i­mum avail­abil­ity of en­gines.

AIR LAUNCHED MINIA­TURE DRONES. What’s small, fast and launched from fighter jets? Not mis­siles, but a swarm of drones. On Jan­uary 10, 2017, US mil­i­tary of­fi­cials had an­nounced that they car­ried out their largest ever test of a drone swarm re­leased from fighter jets. In the tri­als, three F/A-18 Su­per Hor­nets re­leased 103 Perdix drones which then com­mu­ni­cated with each other and went about per­form­ing a se­ries of for­ma­tion fly­ing ex­er­cises that mimic a sur­veil­lance mis­sion.

But the swarm doesn’t know how ex­actly it will per­form the task be­fore re­lease. Perdix drones are not pre-pro­grammed syn­chro­nised plat­forms. These are a col­lec­tive or­gan­ism, shar­ing one dis­trib­uted brain for de­ci­sion-mak­ing and adapt­ing to each other like swarms in na­ture. Be­cause ev­ery Perdix com­mu­ni­cates and col­lab­o­rates with ev­ery other Perdix, the swarm has no leader and can grace­fully adapt to drones en­ter­ing or ex­it­ing the team.

Re­leas­ing drones from a fast-mov­ing jet isn’t straight­for­ward, as high speeds and tur­bu­lence can cause dam­age. But the Perdix drone, orig­i­nally de­vel­oped by MIT re­searchers and named af­ter a Greek myth­i­cal char­ac­ter who was turned into a par­tridge, is now in its sixth it­er­a­tion and able to with­stand speeds of Mach 0.6 and tem­per­a­tures of -10 °C dur­ing re­lease.

At roughly a few me­tres in length, the mis­sile-shaped drones that will make up the US Grem­lin UAV Pro­gramme are built to be launched from the Lock­heed C-130 Her­cules trans­port air­craft. Once the mis­sion is com­plete, the C-130 fishes the Grem­lins out of the air us­ing a spe­cial cap­ture de­vice and car­ries them home where ground work­ers pre­pare them for their next mis­sion within 24 hours. The pro­gramme pro­vides an af­ford­able so­lu­tion to con­duct air com­bat op­er­a­tions in the grow­ing Anti-Ac­cess/AreaDe­nial (A2AD) en­vi­ron­ment. Grem­lins are small and af­ford­able un­manned ve­hi­cles that can be de­ployed in higher risk sce­nar­ios un­suit­able for manned air­craft.

AI IN PRE­DIC­TIVE MAIN­TE­NANCE. The loud­est in­dus­try buzz has been about us­ing big data and AI for pre­dic­tive main­te­nance or turn­ing un­sched­uled events into sched­uled ones by fore­cast­ing fail­ures. But sur­prise events still oc­cur and AI can also help trou­bleshoot them faster and more ef­fec­tively.

Any tool that en­ables pre­dic­tive main­te­nance also helps trou­bleshoot­ing, as it of­ten points to causes of likely fail­ures. Pre­dic­tive an­a­lyt­ics can help op­ti­mise main­te­nance plan­ning and ca­pac­ity by re­duc­ing the need for rou­tine main­te­nance and only trig­ger­ing re­pairs when needed – help­ing in­crease fleet avail­abil­ity by up to 35 per cent and re­duce labour costs by 10 per cent. AI is help­ing bring this to re­al­ity by us­ing data from in-ser­vice air­craft to pre­dict po­ten­tial is­sues. These al­go­rithms are learn­ing to pre­dict de­lays and faults, giv­ing air­lines, air­ports and MROs a bet­ter chance of avoid­ing them.

Boe­ing is now test­ing aug­mented re­al­ity on smart glasses to show me­chan­ics hands-free, in­ter­ac­tive 3D wiring di­a­grams, rather than forc­ing them to view two-di­men­sional, 20-ft-long draw­ings and re­tain that in­for­ma­tion while do­ing re­pairs.

CASE STUD­IES. Case­bank Tech­nolo­gies has been help­ing air­lines with di­ag­nos­tics since 1999 and now sup­ports 10,000 air­craft op­er­ated by 300 com­pa­nies, in­clud­ing some very large air­lines. Case­bank has two ba­sic ap­pli­ca­tions, Spot­Light and Chron­icX. Spot­Light stores data on symp­toms, causes and solutions of com­po­nent fail­ures. Then, through di­ag­nos­tic rea­son­ing, it rec­om­mends op­ti­mal trou­bleshoot­ing steps. The data come from both OEM man­u­als and customer ex­pe­ri­ences in fixing past de­fects. The ap­pli­ca­tion rec­om­mends di­ag­nos­tic steps, but not re­pair in­struc­tions. These are given in the OEM manu



als to which Spot­Light links. Case­bank’s sec­ond ap­pli­ca­tion, Chron­icX, de­tects and man­ages re­cur­ring de­fects, ranks chronic prob­lems and high­lights new trends in de­fects. It uses nat­u­ral lan­guage pro­cess­ing to in­ter­pret un­struc­tured data from pi­lot and main­te­nance records and then spots the clus­ters of re­cur­ring de­fects.

Cog­ni­tive com­put­ing can en­able me­chan­ics to see, based on past ex­pe­ri­ence, which trou­bleshoot­ing steps are most likely to fix a prob­lem. Korean Air be­gan us­ing IBM’s Wat­son-pow­ered tools for cog­ni­tive com­put­ing about four years ago. The car­rier started with a sin­gle fleet, but soon ex­tended the so­lu­tion across all its air­craft. IBM solutions work best for air­lines that per­form their own re­pairs be­cause they have ex­ten­sive re­pair data. Ar­ti­fi­cial in­tel­li­gence is mak­ing avi­a­tion op­er­a­tions and main­te­nance more ef­fi­cient and ef­fec­tive.

Air­bus is tak­ing proac­tive steps to im­prove per­for­mance and re­li­a­bil­ity in the area of air­craft main­te­nance. It is do­ing this by mi­grat­ing his­tor­i­cal main­te­nance in­for­ma­tion from air­craft and fleets to a cloud-based data repos­i­tory known as Sky­wise. Air­bus is also in­stalling sys­tems on each air­craft to col­late and record thou­sands of data pa­ram­e­ters in real-time. Af­ter each flight, this data is up­loaded to Sky­wise to be an­a­lysed and to en­able main­te­nance pre­dic­tions for the fu­ture. The Sky­wise an­a­lyt­ics and AI sys­tem used by Air­bus alerts aero­space op­er­a­tors of pre­dic­tive main­te­nance needs and time­lines so they can take proac­tive steps that en­able them to side­step main­te­nance is­sues be­fore they ap­pear. Sky­wise works in con­junc­tion with on­board di­ag­nos­tic sys­tems which can also gen­er­ate an ‘alert’ while the air­craft is in flight, trans­mit­ting de­tails of the prob­lem to the air­line’s tech­ni­cal ground staff be­fore land­ing. The trans­mis­sion sys­tem which uses VHF ra­dio and/or satel­lite com­mu­ni­ca­tions, is called Air­craft Com­mu­ni­ca­tion Ad­dress­ing and Re­port­ing Sys­tem (ACARS).

COM­MER­CIAL TRENDS HIT­TING AVI­A­TION: DIG­I­TAL TWINS. De­spite longer-last­ing air­craft, more durable en­gines and in­no­va­tions in main­te­nance tech­niques, re­cent re­search has shown main­te­nance spend­ing con­tin­ues to in­crease. How can air­lines keep air­craft in the air while re­duc­ing main­te­nance costs?

Dig­i­tal twins, a state-of-the-art method of mon­i­tor­ing en­gines when in use, will help air­lines achieve these aims. A dig­i­tal twin refers to a vir­tual replica of a phys­i­cal as­set like an air­craft en­gine, which can dis­play how the en­gine is run­ning to engi­neers on the ground. These can then be linked to IT sys­tems to help stream­line and op­ti­mise main­te­nance pro­cesses and op­er­a­tional avail­abil­ity.

To make this hap­pen, engi­neers com­pile thou­sands of data points spe­cific to each as­set dur­ing the de­sign and man­u­fac­tur­ing phase of the en­gine. These are then used to build a dig­i­tal modal that tracks and mon­i­tors an as­set in real-time, pro­vid­ing es­sen­tial in­for­ma­tion through­out an as­set’s life­cy­cle such as en­gine tem­per­a­ture, pres­sure and air­flow rate. By im­ple­ment­ing dig­i­tal twins and cre­at­ing a vir­tual model of the as­set, or­gan­i­sa­tions can re­ceive early warn­ing, pre­dic­tion and even a plan of ac­tion by sim­u­lat­ing “what-if ” sce­nar­ios based on weather, per­for­mance, op­er­a­tions and other vari­ables, help­ing keep air­craft longer in ser­vice.

GE helped de­velop the world’s first dig­i­tal twin for an air­plane’s land­ing gear. Armed with this sort of data, engi­neers and MROs can com­pare data gath­ered by sen­sors on the as­set to that of its dig­i­tal twin, which can be put through the same paces the en­gine ex­pe­ri­ences as it takes off, flies through dif­fer­ent types of weather and un­der­goes reg­u­lar wear and tear. If the two data sets don’t match, then a re­quest can be put in for the en­gine to en­ter ser­vic­ing.Com­pa­nies that in­vest in dig­i­tal twins will see a 30 per cent im­prove­ment in cy­cle times of crit­i­cal pro­cesses, in­clud­ing main­te­nance. In­dus­try ex­pects to see more ben­e­fits as the tech­nol­ogy ma­tures.

AI AS A SER­VICE. As dig­i­tal­i­sa­tion trans­forms busi­ness models, the ap­pli­ca­tion of ad­vanced an­a­lyt­i­cal meth­ods from AI will no longer just be good to have as it will soon be busi­ness crit­i­cal. One of the main chal­lenges fac­ing AI adopters is that stor­ing and analysing vast quan­ti­ties of data can over­whelm IT sys­tems. The next gen­er­a­tion of cloud solutions are here to help process data from pre­dic­tive main­te­nance to in-flight per­for­mance and ag­ing of the air­craft can be bet­ter tracked and un­der­stood.

Soft­ware as a ser­vice so­lu­tion are help­ing drive new ef­fi­cien­cies into com­mer­cial avi­a­tion op­er­a­tions par­tic­u­larly for busi­ness needs such as line main­te­nance ex­e­cu­tion and plan­ning. Pre­vi­ously, air­lines and MROs were con­cerned about the amount of phys­i­cal hard­ware they might need to adopt new tech­nolo­gies, but the trans­for­ma­tion into a SaaS/mo­bile en­vi­ron­ment us­ing tablets or de­vices and elim­i­nat­ing the cost of pur­chas­ing and man­ag­ing on premise tech­nol­ogy, is prov­ing to be at­trac­tive. Cloud-based mo­bile solutions can be rolled out to the work­force with no phys­i­cal in­stal­la­tion re­quired.

DRONES AND FU­TURE TRA­JEC­TORY OF AI. Drones as an au­ton­o­mous in­spec­tor is gain­ing ground. Typ­i­cal vis­ual in­spec­tions of com­mer­cial air­craft can take up to six hours. Drones have the po­ten­tial to cut this time dra­mat­i­cally while of­fer­ing greater ac­cu­racy of checks, free­ing up en­gi­neer time, re­duc­ing main­te­nance costs and im­prov­ing safety. Ini­tial drone sys­tems have al­ready been used to en­hance vis­ual checks made by engi­neers. Low-cost car­rier easyJet has been test­ing drones for fuse­lage in­spec­tions and is look­ing to fully im­ple­ment the so­lu­tion for hail and light­ning strike dam­age. Work­ers would still con­trol the flight of the drone, but by us­ing vis­ual pro­cess­ing al­go­rithms com­bined with en­ter­prise IT sys­tems. This means the drone can send work or­ders straight to the main­te­nance crew as soon as a fault is iden­ti­fied. But chal­lenges re­main. Drones must re­ceive FAA ap­proval for both out­door and in­door flights. FAA Part 107 re­quires un­manned air­craft op­er­a­tors to en­sure that air­craft and con­trols are fit for safe op­er­a­tion prior to flight. Re­gional reg­u­la­tions that change from coun­try to coun­try must also be con­sid­ered, as do op­er­a­tional com­pli­ca­tions such as se­cu­rity safe­guards, com­mu­ni­ca­tion with on­go­ing air traf­fic and air­port au­thor­ity ap­proval to make sure drones are used safely.

Air­lines, MROs and other par­ties are con­stantly look­ing to make ma­jor im­prove­ments in op­er­a­tional pro­cesses and, al­though these tech­nolo­gies may be at the start of their avi­a­tion life­span, the com­mer­cial avi­a­tion in­dus­try is fully aware of the ben­e­fits they will bring.



At just 4 cm long, these pro­to­type SWARM ro­bots are pi­o­neer­ing our In­tel­li­gent En­gine vi­sion. Col­lab­o­ra­tive and en­abled with Ar­ti­fi­cial In­tel­li­gence, these tiny ro­bots will trans­form the fu­ture of en­gine main­te­nance.

Rolls-Royce SWARM Ro­bot

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.