DEVELOPMENT AND EVALUATION OF STRATEGIES FOR PILOTING ASSISTANCE FOR LIGHT AIRCRAFT

Transcription

1 7 TH INTERNATIONAL CONGRESS OF THE AERONAUTICAL SCIENCES DEELOPMENT AND EALUATION OF STRATEGIES FOR PILOTING ASSISTANCE FOR LIGHT AIRCRAFT Antonio Rafael da Sila Filho, Francisco Ribeiro Fonseca Reis, Guilherme André Santana, Igor Machado Malaquias, Paulo Henriques Iscold Andrade de Olieira Departamento de Engenharia Mecânica Centro de Estudos Aeronáuticos Uniersidade Federal de Minas Gerais Brasil Keywords: facilitated airplane, fly-by-wire, light aircraft, pilot assisted Abstract The objectie of this work is to deelop and study seeral facilitated airplane strategies to be used in light aircrafts in order to make flight more intuitie, improing learning time and flight safety. A flight simulator will be implemented to test studied strategies with arious users haing different piloting skills. For each strategy it will be gien a score considering user s feeling as well as his capability tofollow a predetermined path. 1 Introduction Recently, growth of market of small aircraftshas shown a popularization of so called light aircrafts [1]. Most of time, those airplanes are personal airplanes used for leisure and short traels. While in general aiation most pilots are professional pilots, personal airplanes pilots are usually owners of m and most of time are ineperienced pilots. This fact decreases flight safety especially in situations that demand attention and a higher leel of piloting effort such as piloting, naigating and communicating with ground stations simultaneously. To sole this issue researches are being conducted to deelop facilitated airplanes systems for light aircrafts, based in fly-by-wire systems [], in order to make flight safer by reducing piloting effort and so making it easier. This work will study seeral facilitated airplane strategies to permit longitudinal control of airplane based directly in its trajectory instead of its attitude. For so, automatic controllers strategies hae been chosen based in reference ariables directly linked to airplane longitudinal moement. The specific objecties of this work are: Design and implementation of a longitudinal flight simulator in order to test proposed strategies. Ealuate suitability of proposed strategies through flight simulations inoling people with different piloting skills and knowledge. First, a dynamic model of aircraft chosen for flight simulator, CB-10 Triatlhon [3], will be presented. The computational implementation of flight simulator and definition of studied strategies as well as ir adjustment within simulator will be presented net. At last, used ealuation methodology and most important results obtained will be eposed. Airplane Dynamic Model Fig.1 presents dynamic model of simulator airplane. The CB-10 Triatlhon was chosen because it is a light aircraft that in near future will become a base for flight tests and facilitated flight systems researches conducted by Centre for Aeronautical 1

2 ANTONIO RAFAEL DA SILA FILHO, FRANCISCO RIBEIRO FONSECA REIS, GUILHERME ANDRÉ SANTANA, IGOR MACHADO MALAQUIAS, PAULO HENRIQUE ISCOLD ANDRADE DE OLIEIRA Studies of Gerais. Federal Uniersity of Minas Fig. 1.. Dynamic Model Based on reference [], airplane equations of motion will be presented. The model features three degrees of freedom: i) displacemen nt along ais, ii) displacemen nt along y ais and iii) longitudinal rotation oer z ais. Therefore, re are si state ariables for considered model described as follows: They can also be written as: y 1 L sin D cos m LT sin t T( Pp) cos 1 L cos D sin 5 m LT cos t T ( Pp) sin W 1 L A cos DA sin 6 J LT T cos T M y y q (1) () Where: TT ( ( = + u NT L = 1 r S C ( ) R w L( a) D = 1 r S C ( ) R w D( a) M = 1 r S c C ( ) R w M( a) T = C r n. D é u y y g = arctga - ê ë a = q- g L R T = 1 r S C ( ) T = = + u cos + C LT a ( u ) ) ( ) ( u ú ) 3 ) sen ( RT T = t TT LT at ; d - - u cos + u u = ( 6 = atg T 3 NT T de w = a d a g + = - a ( ) RT ( - u ) ( - 5 y ) ) = é( a + i ) a ù é êë d a ù T t 1úû + êë úû 5 y u sen R T + ú û y RT TT ) ù ú w (3)

3 DEELOPMENT AND EALUATION OF STRATEGIES FOR PILOTING P ASSISTANCE FOR LIGHT AIRCRAFT The thrust generated by engine considering propeller efficiency was calculated as in reference [5], making it possible to obtain a table relating propeller thrust coefficient, aircraft speed, percentagee of engine power and engine rotation. The presented state model has two control ariables, eleator deflection d directly related to horizontal tail lift and engine power percentage Pp. Thus, se are two control ariables that will be used in simulator. 3 Flight Simulator Implementation The simulator was implemented using numerical software Matlab through Simulink interface. The toolbo AeroSim was used to communicate with flight simlulator FlightGear that was used as a graphical interface. The integrationn enoling Matlab software with FlightGear graphical interface proides a better interaction with user compared to graphic response of simulator. The AeroSim toolbo has seeral useful tools which were used to t design simulator instruments panel. Basic instruments usually present in light airplanes were chosen, and also some ors to measure ariables related to longitudinal moement such as: attack angle, pitch angle, elocities angle and load factor as shown in Fig.. pointer ( white w one) shows instantaneous ariable alue while or ( red one) shows eference alue that ariable is heading to. Facilitatedd Flight Strategies As it was shown before, dynamic model of airplane has two control ariables: d and Pp. Inn a conentional airplane those ariables are set directlyy by pilot trough stick and e throttle in order to control attitude of t airplane. Fig. 3 presents a simplified scheme of a conentional airplane longitudinal control. c Fig. 3. ConentionaC al Control System Applying pre-established control laws using electronic mechanisms to manage command-attitude-trajectory interface, it will be possible for pilot to control directly airplane trajectory as airplane automatic control system performs all necessary calculations. The facilitated flight strategies are programmed in controllers with feedback from commands referencee output alues. In that way, through pre-established control laws, it is possible too determinee best configuratio on for airplane in orderr to fly desired path Fig.. Fig.. Instruments Panel Two pointers were used in each instrument due to fact that studied strategies are based in reference alues of longitudinal moement related ariables. One Fig.. Facilitated F Flight Strategies Implementation 3

4 ANTONIO RAFAEL DA SILA FILHO, FRANCISCO RIBEIRO FONSECA REIS, GUILHERME ANDRÉ SANTANA, IGOR MACHADO MALAQUIAS, PAULO HENRIQUE ISCOLD ANDRADE DE OLIEIRA Because of that, instead of controlling directly engine power percentage and eleator s angle through throttle and stick, pilot will control two or ariables related to airplane longitudinal moement. The automatic controller will n calculate necessary power and eleator s angle in orderr to reach alues specified by pilot for new controlled ariables. Thus it is necessary to select ariables related to airplane longitudinal moement that are propitious to be implemented in facilitated flight strategies. These are chosen candidates: Speed []; Altitude [h]; Rate of climb [ROC]; Pitch angle [q ]; Angle described by ertical speeds [ g horizontal and ] Each combination of aboe ariables taken two by two willl define a strategy s to be used. Table 1 summarizes chosen strategies for this work. The reference columns represent ariables to be set by pilot while actuation columns show ariables that will be modifiedd by actuating eleator or by changing engine power in order to achiee set points proided by pilot. Table 1 Implemented Strategies Estrategy Reference Actuation Stick Throttle Eleator Engine h h ROC ROC H ROC h ROC 05 q q 06 q q 07 g g 08 g g Along this work w strategies that t use engine powerr to modify y airplane speed will be called direct strategies while ones that use eleator for e same purpose will be called crossedd strategies.. Haing defined all strategies to be used it is necessary to implement controllers to proide automatic a control of t reference ariables for each strategy. In this work it will be used proportional, integratie, deriatie (PID) controllers. The net n step consists in implementing controllers into each strategy. The set points chosen by e pilot for controlled ariables are comparedd to ir actual alues and error is applied to controller input. The PID will n use e eleator or engine power (depending onn strategy) in orderr to make error be equal to zero. A bloc diagram of implementation is proided by Fig. 5. As each strategy actss into twoo ariables, PID actuated by t stick will be called PID 1 and or, actuated by e throttle, PID. Fig. 5. Implementation of PID into Strategies All aboe definitions are required to adjust gain alues of each strategy PIDs. Particular effort must be made in order to achiee a similar response characteristic for each strategy.. It must be noted that objectie of this work is not to deelop optimumm controllers but to compare studied strategies. Thus it is important that gains adjustment do not interfere in i ealuation process. The controller ss adjustment will be diided in three steps: Although it seems more natural that power throttle should control airplane speed while stick should control or chosen ariable, this restriction will nott be imposed. Initial gain aluess selection. Definition of a characteristic response from controllers with feedback (objectie function).

5 DEELOPMENT AND EALUATION OF STRATEGIES FOR PILOTING P ASSISTANCE FOR LIGHT AIRCRAFT Adjustment of controllers gains in orderr to achiee characteristic response defined. The indirect method of Ziegler Nichols [ 6] was used to determine initial gain alues in order to feature typical characteristics off a PID controller, stable, damped and with no steady error. Net, performance indeess related to response with feedback must be elaborated and studied. The response must be stable, s withh a short stabilization time and with no steady error. An objectie function will be created for that purpose as a function of desired performance e indees. Last, using optimization algorithms, controllers gains must be refined so that system response achiees epected indees. As shown before, each strategy will hae two different control ariables. It is important to note that optimization of each ariable s PID ( PID 1 and PID ) must be performed considering or ariable s PID optimization since one influences or performance e. 5 Tests First of all, a reference trajectory t was defined after studying simulated airplane performance e in orderr to represent a high performance e flight. The trajectory consists inn a series of patches corresponding to leeled flight, climbing and diing. All oer trajectory airplane speed and position can be computed for furr analysis. Fig. 6 presents t predefined flight trajectory Speed 50m/s 6 8 Distance [km] Speed 60m/s Fig. 6. Reference Trajectory 10 1 For a better identification of reference trajectory, circles to be flown through by user hae beenn implemented in FlightGear interface. Each part of t trajectory is identified by b a different color so user can easily know where w he is flying. Fig. 7 features a screenshot of simulator graphic interface with t reference circles implemented. Obiously, it would be no need of se circles if strategies were supposed to be b implemented in a real airplane. Fig. 7. Simulator s Graphic Interface Screenshot The quality off each strategy was measured by t capability of user to stick to referencee trajectoryy and reference speed. Each user was asked too fly with each of eight strategies and withh no strategy at all in a random order. The strategies, including no strategy condition, were n ealuated in a subjectie andd an objectie ways. The subjectie ealuation was conducted using Cooper-Harperr scale [7]. That scale is widely used to classifyy an airplane s control qualities, 1 representingg good control qualities with no improements required and, 10 representing an incontrollable airplane. Each user, after each flight, was asked to gie a score to flown strategy based on hiss feeling and comfort following Cooper-Harper scale. The desired performance is reached when user is able to control airplane following established path, p without physical and mental efforts. The acceptable performance is when minor non continuous efforts are required for same purpose. 5

6 ANTONIO RAFAEL DA SILA FILHO, FRANCISCO RIBEIRO FONSECA REIS, GUILHERME ANDRÉ SANTANA, IGOR MACHADO MALAQUIAS, PAULO HENRIQUE ISCOLD ANDRADE DE OLIEIRA The objectie ealuation was conducted by computing all flight, altitude and speed instant errors calculated as follows: Speed Error = - () flight flight reference Hight Error = h -h (5) reference corresponds to: A control system that can not ensure a suitable performance with a tolerable amount of work. An improement is mandatory. The systems features serious deficiencies (most users did not hae piloting eperience). Table Strategies score (subjectie) The total errors, called accumulated errors, were n calculated by integrating instant errors all along simulated flight. Speed Acumulated Error = final 0 0 Speed Error d Hight Acumulated Error = final Hight Error d (6) (7) No strategy Reference (Speed and Altitude) Str. 1 (Crossed) Str. (Direct) Reference (Speed and Rate of Climb) Str. 3 (Crossed) Str. (Direct) Reference (Speed and Pitch Angle) Str. 5 (Crossed) Str. 6 (Direct) Reference (Speed and Speeds Angle) Str. 7 (Crossed) Str. 8 (Direct) According to that error formulation, error represents how much user flown out of reference trajectory and speed. Two by-pass patches were added at beginning of flight and at transition speed in order to gie user some time to trim airplane and get familiarized with controls. 5 Results 5.1 Subjectie Ealuation The strategies were tested by thirty four users and Table presents mean score gien to each strategy following Cooper-Harper scale as well as standard deiation (S. D.). As can be noted from Table, seen of eight studied strategies, according to users opinion, showed an improement in flight quality compared to no strategy condition. Strategy 5 (Crossed, Reference: Speed and Pitch Angle) was only that did not presented an improement. It must be noted that mean score for no strategy condition was 7., which according to Cooper-Harper scale Mean S. D According to users, strategies (Direct, Reference: Speed and Rate of Climb) and 8 (Direct, Reference: Speed and Speeds Angle) achieed best improement to flight quality with a score inferior to 3. In Cooper- Harper scale that corresponds to: Satisfactory. No improements needed. Negligible deficiencies. It was classified between reasonable and good. It is also interesting to note that among all strategies direct ones presented a better score than crossed ones. Only strategy 8 that scored. for direct mode (good) had a crossed mode that was rated below (still in no improement needed category). Howeer, it is necessary to keep in mind that those results are only representatie for a flight trajectory with characteristics similar to studied one. 6

7 DEELOPMENT AND EALUATION OF STRATEGIES FOR PILOTING ASSISTANCE FOR LIGHT AIRCRAFT 5. Objectie Ealuation Fig. 8 features mean cumulated error referring to flown flight altitude difference relatiely to reference trajectory. The accumulated error was normalized using no strategy condition as a reference. Strategy 1 (Crossed, Reference: Speed and Altitude) presented a worsening of 60% relatie to no strategy condition and achieed worst performance of all strategies. In or hand strategy 8 (Direct, Reference: Speed and Speeds Angle) presented an improement of almost 80%, achieing best performance of all strategies. more than 80% of improement. Once more, strategy 8 (Direct, Reference: Speed and Speeds Angle) was best rated strategy. Following, a short analysis comparing best strategies ( and 8) to no strategy condition will be performed. 5.3 Flown Trajectory Analysis Fig. 10 presents users trying to fly following reference trajectory with no facilitated flight strategies. Whereas y were flying through indicated black lines error was not computed. It was considered a transition region as mentioned earlier in this article. No Strategy Altitude (Crossed) Altitude (Direct) Rate of Climb (Crossed) Rate of Climb (Direct) Pitch Angle (Crossed) Pitch Angle (Direct) Speeds Angle (Crossed) Speeds Angle (Direct) Accomulated Altitude Error Fig. 8. Mean Accumulated Error for Speed (normalized) No Strategy Altitude (Crossed) Altitude (Direct) Rate of Climb (Crossed) Rate of Climb (Direct) Pitch Angle (Crossed) Pitch Angle (Direct) Speeds Angle (Crossed) Speeds Angle (Direct) Accomulated Speed Error Fig. 9. Mean Accumulated Error for Altitude (normalized) Fig. 9 features mean cumulated error referring to flown flight speed difference relatie to reference speed. It is interesting to note that all strategies, een crossed ones, achieed an improement of more than 60% regarding no strategy condition. Again direct strategies obtained a better score than crossed ones. All direct strategies achieed Proposed Trajectory Flight Trajectory Fig. 10. No Strategy Proposed and Flown Trajectories The results concerning strategies and 8 are shown in Fig. 11 and Fig. 1. The improements are outstanding. The strategies clearly facilitated trajectory control by making it possible for users to stick close to reference trajectory Proposed Trajectory Flight Trajectory Fig. 11. Strategy Proposed and Flown Trajectories 7

8 ANTONIO RAFAEL DA SILA FILHO, FRANCISCO RIBEIRO FONSECA REIS, GUILHERME ANDRÉ SANTANA, IGOR MACHADO MALAQUIAS, PAULO HENRIQUE ISCOLD ANDRADE DE OLIEIRA Proposed Trajectory Flight Trajectory Fig. 1.. Strategy Proposed and Flown Trajectories Speed [m/s] Reference Flown No Computation Stall Fig. 1. Strategy Proposed and Flown Speeds 5. Flown Flight Speed Analysis Fig. 13 features flown flight speed with no piloting assistance strategies and reference flight speed. It is important to note that some users had great difficulties controlling flight speed and een let it decrease below stall angle compromising flight safety. Speed [m/s] Reference Flown No Computation Stall Fig. 13. No Strategy Proposed and Flown Speeds Fig. 1 and Fig. 15 show flown flight speed for assisted flights using strategies and 8. Once again assisted flights performed much better than non assisted flight. Especially, no user let airplane stall, which characterizes a safety improement. Speed [m/s] Reference Flown No Computation Stall Fig. 15. Strategy 8 Proposed and Flown Speeds 5.5 Eperienced Users and no Eperienced Users Comparison A comparison regarding eperienced and no eperienced users flights can help alidating hyposis that assisted flight strategies improe airplane flight qualities. As an eample, two of users will be compared: An eperienced pilot and a user with no piloting skills and non aeronautical knowledge. Fig. 16 presents a comparison of flown trajectory of two compared users. As epected eperienced pilot stuck closer to predetermined trajectory while or user did not performed so well, particularly at first climbing path. 8

9 DEELOPMENT AND EALUATION OF STRATEGIES FOR PILOTING ASSISTANCE FOR LIGHT AIRCRAFT Reference No Computation Pilot No Pilot Speed [m/s] Reference No Computation Piloto No Pilot Fig. 16. Trajectory Comparison No Flight Assistance Fig. 18. Flight Speed No Flight Assistance Fig. 17 features a comparison of same users flying with control strategy 8. It is clearly seen that both users had a much closer performance. It can also be noted that eperienced pilot did not lose any performance using assisted flight strategy Reference No Computation Piloto No Pilot Fig. 17. Trajectory Comparison Strategy 8 The same analysis was performed for flown flight speed. Fig. 18 shows that both users had some issues controlling flight speed but specially ineperienced user. Fig. 19 shows compared users flying with assisted flight strategy 8. Both users controlled flight speed much easier and achieed a comparable performance. The comparisons analysis suggest that implementation of facilitated flight strategies can bridge gap between eperienced and ineperienced pilots. Speed [m/s] Reference No Computation Piloto No Pilot Fig. 19. Flight Speed Comparison Strategy 8 5 Conclusion Along this work, assisted flight strategies were deeloped and studied based on airplane trajectory control instead of usual airplane attitude control. The studied strategies were ealuated by thirty four users haing different flight skill and aeronautical knowledge. The ealuation was carried out using a flight simulator deeloped and implemented for that specific purpose. Eight assisted longitudinal flight strategies were ealuated haing as reference ariables: altitude, rate of climb, pitch angle and speeds angle. For each chosen combination of those reference ariables re were proposed two modes: direct mode (speed ariation controlled by engine power percentage) and crossed mode (speed ariation controlled by eleators deflection). For all combinations direct modes achieed better performance than crossed modes. In some cases crossed strategies induced a worst performance comparing to no 9

10 ANTONIO RAFAEL DA SILA FILHO, FRANCISCO RIBEIRO FONSECA REIS, GUILHERME ANDRÉ SANTANA, IGOR MACHADO MALAQUIAS, PAULO HENRIQUE ISCOLD ANDRADE DE OLIEIRA strategy condition. That can be seen for instance in Fig. 8 where strategies 1 (Crossed, Reference: Speed and Altitude) and 5 (Crossed, Reference: Speed and Pitch Angle) clearly were oer performed by no strategy condition. Strategy 8 (Direct, Reference: Speed and Speeds Angle) was most performing strategy for this work. According to users opinion that strategy achieed a score of. in Cooper Harper s scale and was classified as: Satisfactory. No improements needed. Negligible deficiencies. Desired Performance Achieed with no Pilot Efforts. Strategy was second more performing strategy scoring 3 in Cooper Harper scale and classified as: Satisfactory. No improements needed. Unpleasant deficiencies. Desired Performance Achieed with minimum pilot efforts. Strategies 8 and are thus good candidates to be implemented in flight assistance systems. Last, a comparison between an eperienced pilot and an ineperienced user with no flight skills showed that ir performance could be brought to same leel using assisted flight strategies. References [1] Finotti J. Jet Sets NASA s plan to reolutionize how Floridians fly in future Florida Trend The Magazine of Florida Business SATS, 00 [] Tomczyk A. Proposal of eperimental simulation method for handling qualities ealuation Department of Aionics and Control Systems, Faculty of Mechanical Engineering and Aeronautics, Rzeszo w Uniersity of Technology, Rzeszo w, Poland; Aircraft Engineering and Aerospace Technology: An International Journal, 008. [3] Barros C P. Uma metodologia para o desenolimento de projeto de aeronaes lees subsônicas Uniersidade Federal de Minas Gerais Escola de Engenharia Programa de Pós-Graduação em Engenharia Mecânica, 001. [] Olieira P H I A. Otimização de trajetórias de ôo de distância em planadores - Tese de Doutorado em Engenharia Mecânica - Uniersidade Federal de Minas Gerais, UFMG, Brasil, 00. [5] Hartman E P. Biermann D. The aerodynamic characteristics of full-scale propellers haing, 3, and blades of Clark y and R.A.F. 6 airfoil sections. NACA Report N 60, 1938 [6] Ogata K. Engenharia de Controle Moderno, Quarta Edição, Segunda Reimpressão. Editora Pearson Prentice Hall, 006. [7] Harper R P JR. Handling qualities and pilot ealuation, (Calspan Corp., Buffalo, NY) & Cooper, G. E., (G. E.Cooper Associates, Saratoga), CA Journal of Guidance, Control, and Dynamics ol.9 no.5 (515-59), 1986 Contact Author Address Antonio Rafael da Sila Filho arafael@ufmg.br Francisco Ribeiro Fonseca Reis chicorfr@yahoo.com.br Igor Machado Malaquias imalaquias@yahoo.com.br Guilherme André Santana guilhermeandresantana@gmail.com Paulo Henriques Iscold Andrade Olieira iscold@ufmg.br Copyright Statement The authors confirm that y, and/or ir company or organization, hold copyright on all of original material included in this paper. The authors also confirm that y hae obtained permission, from copyright holder of any third party material included in this paper, to publish it as part of ir paper. The authors confirm that y gie permission, or hae obtained permission from copyright holder of this paper, for publication and distribution of this paper as part of ICAS010 proceedings or as indiidual off-prints from proceedings. Acknowledgements The authors would like to acknowledge FAPEMIG (Fundação de Amparo à Pesquisa do Estado de Minas Gerais) for financial support, and CNPq (Conselho Nacional de Desenolimento Científico e Tecnológico - "National Counsel of Technological and Scientific Deelopment") for support during project. 10