You are on page 1of 5

Robustness Analysis for Rotorcraft Pilot Coupling

with Helicopter Flight Control System in Loop


CHEN Junfeng, LI Wuzhou, SUO Wenchao, WANG Zegang ,XU Penghui, WANG Weilong
Department of Helicopter Engineering, Army Aviation Institute
Beijing, China
cjf_hn0227@126.com, liwuzhou@163.com, suowenchao@163.com, zegang@163.com, penghui@163.com ,weilong@163.com

Abstract—Robustness analysis method is proposed for by the pilot’s passive control is in high frequency range of 2-
rotorcraft pilot coupling with helicopter flight control system in 8Hz.
loop. Combining with the lateral identification model of BO-105
helicopter, McRuer’s pilot model, and the designed stability As for RPC research, refs. [9] analyzes the unsteady
augmentation system, frequency domain model is established for phenomenon of vertical cone mode, which is caused by the
rotorcraft pilot coupling analysis. ȝ analysis method and passive control of the pilot based on Mayo’s biodynamic
performance specifications in ADS-33E are adopted to compare feedthrough modeling. Refs. [10] and [11] analyze the vertical
robust performances of the closed loop, furthermore, the worst oscillation of the helicopter in time domain combining
case in uncertain model set is studied using robust performance McRuer’s crossover model and the Mayo’s biodynamic
detection. Moreover, the properties of root locus, damping ratio feedthrough model. However, Uefs. [9]-[11] have not
and mode frequency are studied from flight control system aspect. considered FCS in closed-loop of the helicopter. Hess analyzes
Feasible measures to improve the robust performance are RPC oscillations based on structural model in different flight
discussed from the controller design consideration. Results show tasks, such as landing, hovering, acceleration/deceleration,
that the analysis method proposed can not only reveal the etc.[12]-[15], which are focused on the time-domain simulation,
physical nature of coupling qualitatively, but also calculate the
lacking of enough discussion about the RPC mechanism.
performances and boundaries quantitatively, which have
theoretical reference significance in flight control system design. The resent researches of the RPC mainly adopt the method
by combining different pilot models with the helicopter
Keywords—rotorcraft pilot coupling; robustness analysis; flight dynamics models to build the RPC closed loop. However, the
control system; pilot model; mode characteristics closed-loop models in references rarely include FCS, although
it is an important role in RPC loop. Hess considers the FCS in
I. INTRODUCTION closed loop modeling, but mainly uses time domain analysis
FCS(Flight Control System) with high gain and high method. In this paper, frequency domain analysis method is
bandwidth has been an important target in advanced helicopter adopted to study the stability and robustness of RPC,
design to get better handling quality. However, FCS design considering the pilot control and FCS. The structure is
needs to concern about the effects when the pilot is in the organized as follows: Section 2 presents RPC model, includes
closed loop, for the reason that the helicopter performances helicopter modeling, the pilot modeling and the FCS modeling.
may be weaken, and even RPC(Rotorcraft Pilot Coupling) is Section 3 proposes robustness analysis methods. Section 4
induced, which is dangerous to flight safety[1][2]. performs case study, and analyzes the RPC robustness. Finally,
Section 5 concludes the paper and points out further work.
In order to analyze the effects of the pilot in the closed loop,
pilot modeling has been researched firstly for many years[3], e.g. II. RPC MODELING
McRuer has established the crossover model[4][5], which
manifests that transfer function of the open loop with the pilot In order to facilitate the analysis, following simplifications
has the frequency characteristics of H (s) Zc <eW s s , ( Zc is the
e are made to the RPC model:
crossover frequency, W e is the equivalent lag time). The gain (1) Single channel of the helicopter model is adopted
is adjusted by pilot to make Zc close to the highest cut-off without considering the couplings. In order to ensure the
authenticity of the simulation, the lateral identification model
frequency. Hess has established the structural model[6][7], which of BO-105 helicopter is used.
uses the visual perception, perception of neuromuscular system
and vestibular system perception to synthesize the feedback (2) Flight task is assumed to maintain the lateral attitude in
signal, by which interactions between pilot and helicopter are near earth hovering, without considering other channels control
described with the double feedback loops. Mayo has and channel couplings
established biodynamic feedthrough model[8], which represents
the interactive coupling behavior between control stick and the (3) Stability augmentation is supposed to be used in
pilot, and it is essentially a passive pilot model. Generally, the hovering, and the control response type of the lateral channel
oscillation caused by the active control of the pilot is in low is RCAH(Rate Command Attitude Hold).
frequency, which is about 1-2 Hz. While the oscillation caused

978-1-4673-8979-2/17/$31.00 ©2017 IEEE 68

Authorized licensed use limited to: Air University. Downloaded on December 27,2022 at 17:04:24 UTC from IEEE Xplore. Restrictions apply.
The closed loop structure of RPC is shown in Fig.1, the neuromuscular delay, which limits the pilot control rate, and its
inner loop with FCS is Gin. Roll rate p and roll angle I are used value usually equals 0.10; k pl is pilot control gain, which
as the controller signals, which correspond to the coefficients needs to be adjusted to guarantee the optimal operation, and it
K p and KI . The pilot adjusts the flight strategy according to r usually ranges from 0~100; TL is the lead compensation time
and I . The control input u is the output of the stability constant, which depends on the pilot’s prediction ( TL z TN ),
augmentation operation between pilot control instructions and and reflects the pilot mental load, it usually ranges from 0~2.5;
the negative feedback input. Meanwhile, d is the external Tl is the time delay of the central information, which reflects
disturbances (such as gust) to the helicopter. the value of pilot’s physical strength load, and it usually ranges
from 0~20; k pl < TL s  1 Tl s  1 is a tunable part of the
I model, which is equivalent to lag-lead compensator. The
closed-loop bandwidth with pilot in loop is usually less than
4rad/s, if there exists an unstable mode frequency which is
higher than this value, it will be difficult for the pilot to control
the helicopter.

C. FCS Modeling
Fig. 1. Closed loop structure of RPC.
In order to build a closed-loop and guarantee that the
RPC modeling of the closed-loop mainly contains the controller design is reasonable, stability augmentation system
helicopter modeling, the pilot modeling and the FCS modeling, is designed in the following form
which are built as follows. k p kd s  1
u <'I K p <'p  KI <'I (3)
A. Helicopter Modeling Ti s  1

According to refs. [16], lateral identification model of the where,


helicopter is of the form
k p kd s
2.62 u > 0.413 , 3.07 @> 0.0696 , 16.2@ e 0.0225 s Kp
I § rad · (1) Ti s  1
G lat 0 >0.277 ˈ2.75@>0.509 ˈ13.7 @>0.0421 ˈ15.8@ ¨© (%) ¸¹ kp
KI
where [] , Z ]  s 2  2]Z s  Z 2 , 1 T  s  1 T , ] is Ti s  1
damping ratio, and Z is undamped natural frequency.
the normal parameters of the controller are k p 36.639 ,
The identification model can not only reflect the direct kd 0.096 , Ti 0.088 .
correspondence of roll angle I and lateral control amplitude of
the stick G lat (percentage), but also specify natural mode The response p, I of the inner loop Gin to the stick input
characteristics, as shown in Table ĉ. rstk is shown in Fig.2.
15

TABLE I. MODES OF BO-105 IN LATERAL DIRECTION 10


rstk/(%)

0
Mode name Damping ratio Structure frequency 0 2 4 6 8 10 12 14

5
p/(deg/sec)

Roll/Flap 0.51 13.7 rad/s 0

Regressive Lag 0.0421 15.8 rad/s -5


0 2 4 6 8 10 12 14

Dutch Roll 0.277 2.75 rad/s 10

5
I/(deg)

B. Pilot Modeling 0

Considering that McRuer’s pilot model is mature and has a -5


0 2 4 6
t/(sec)
8 10 12 14

clear physical meaning, it is adopted in this paper. The transfer


function of the model is as follows[5]:
Fig. 2. Control responses of Gin.
TL s  1 W s
G pt k pl < <e pl (2)
l N
T s  1 T s  1 III. ROBUSTNESS OF THE RPC
In the formula, s is the Laplace operator; W pl is the time As a tool of the robustness analysis, the ȝ analysis is widely
applied to frequency domain analysis[17], which is adopted in
delay which depends on nerve conduction and stimulation, and this paper. Meanwhile, helicopter response ability is measured
it usually ranges from 0.13~0.30; 1 TN s  1 denotes the by the closed-loop bandwidth ȦBW and the phase delay W p ,

69

Authorized licensed use limited to: Air University. Downloaded on December 27,2022 at 17:04:24 UTC from IEEE Xplore. Restrictions apply.
which have been proposed in ADS-33E(Aeronautical Design the pilot is not in closed loop, the response time of the roll
Standard)[18]. The numerical perturbation analysis base on the angle I of the helicopter is longer and the overshoot is larger
characteristics of the closed loop can measure the robustness of considering that the pilot is in the closed loop. Compared with
the RPC. the nominal model, the response performance of stochastic
According to the hypothesis of RCAH, the closed-loop model is poor, which reflects the weakening effect to closed
bandwidth ȦBW and the phase delay W p are defined as loop with pilot in loop.


ZBW min ZBW , ZBWgain phase
(4)
1.4

1.2

') 2Z180
Wp (5) 1

57.3(2Z180 ) 0.8

I/(rad)
0.6
the parameter definitions in the equation are showed in the
0.4
Fig.3.
0.2

0 without pilot model


nominal case
random case
-0.2
0 2 4 6 8 10 12
t/ (seconds)

Fig. 4. Comparison of step response.

The ȝ analysis of the structure singular values with the


parameters in the uncertain ranges is showed in Fig.5. The
maximum of ȝ is 0.66 in the uncertain set of the pilot model,
which reflects that RPC closed loop has good performance
boundary. The robustness performance boundary is [1.51, 1.72],
it assures that the closed loop of the RPC in the uncertain range
is stable.
Fig. 3. Definitions of bandwidth and phase in ADS-33E.
0.8
UpperBound
LowerBound
IV. NUMERICAL ANALYSIS 0.7

A. Robustness with pilot model in uncertain conditions. 0.6

Considering that the pilot’s manipulation is adjusted in real 0.5


P / (Abs)

time according to the flight condition, the pilot model is


parameter dependent. In order to analyze the influence of 0.4

model uncertainty on robust performance, an uncertainty


parameters set of pilot model is assumed in Table Ċ, in which 0.3

the upper and lower boundary values fluctuates from the 0.2
nominal value.
0.1 -1 0 1
10 10 10
Z / (rad/s)
TABLE II. PARAMETER VALUES OF PILOT MODEL
Fig. 5. ȝ analysis.
Parameter Nominal value Bounds
The worst case of robustness analysis is showed in the
TL 0.26 [-50% , +50%] Table ċ. It can be seen that the closed-loop bandwidth ȦBW
Tl 10.00 [-50% , +50%] declines 27.5% compared with that in nominal conditions, and
the phase delay W p increases 79.2%, reaching 0.430s, which is
IJpl 0.20 [0.13,0.30]
the important index to reflect RPC proneness. Meanwhile, the
TN 0.10 0.10
decline of ȦBW and the increase of W p reflect the performance
kpl 9.26 9.26
degradation when the pilot is in the closed loop compared with
The step response of the closed loop system with uncertain
that when the pilot is out of the loop. The comparison of the
parameters is shown in Fig.4. Compared with the situation that
bode diagram in the three conditions is showed in Fig.6.

70

Authorized licensed use limited to: Air University. Downloaded on December 27,2022 at 17:04:24 UTC from IEEE Xplore. Restrictions apply.
TABLE III. COMPARE OF ROBUST PERFORMANCES INDEX C. Measures to improve the performance of RPC loop.
From the view of controller design, a higher kp and kd is
Without Nominal beneficial to improve the bandwidth and to decline the phase
Parameter Worst case delay of the RPC loop, as is shown in Fig.8. However, as have
pilot model case
been analyzed in Č.B, the high kp and kd can also damage the
ȦBW 5.922 rad/s 1.208 rad/s 0.875 rad/s stability of the roll/flap mode. Thus, there need a compromise
IJp 0.062 s 0.240 s 0.430 s between the robustness and performance of RPC loop.
Ȧ180 12.092 rad/s 2.396 rad/s 1.762 rad/s

20

1.6
0
| I/r| /(dB)

1.4
-20

ZBW /(rad/s)
1.2
-40

1
-60
-1 0 1
10 10 10
Z /(rad/s) 0.8

0.2
-200 0.15 60
50
0.1
I/(deg)

40
-400
0.05 30
20
kd 0 10
-600 without pilot model kp
nominal case
worst case
-800
-1 0 1
10 10
Z /(rad/s)
10
(a) Bandwidth

Fig. 6. Comparition of bode plot.

0.32
B. Stability of the mode.
0.3
According changing the controller parameters kp and kd,
0.28
stability of the mode in lateral channel is analyzed. The
W/(seconds)

simulation shows that the roll/flap mode is most sensitive to 0.26

parameters variation. It can be seen from Fig.7 that this mode’s 0.24

real part and the damping become negative with the increase of 0.22

kp, what lead to the instability of RPC loop. Meanwhile, the 0.2
damping of roll/flap mode becomes smaller with the increase 0.18 60
of kd. Therefore, the stability of roll/flap mode needs to be 0.2
0.15 40
50

taken much care in RPC controller design. 0.1 30


0.05 20
0 10
kd kp
15
]=0.2 ]=0.1
]=0.3
]=0.4
Unstable
(b) Phase delay
14

Fig. 8. Influence of controller parameters to closed loop


13 performances.
kp
Im/(rad/s)

12 V. CONCLUSION
11
A tractable model is offered for the RPC frequency-domain
analysis, and the robustness analysis method is proposed to
10
study RPC loop, which combines the classical stability analysis,
with kd decreased: k d=0.064
nominal case: k d=0.096 ȝ analysis and mode characteristics analysis. The simulation
9
with kd increased: kd=0.128
results manifest that the performances of the closed loop in
-7 -6 -5 -4 -3 -2
Re/(rad/s)
-1 0 1 2 3
lateral are weakened with pilot in loop, moreover, the roll/flap
mode is prone to become unstable with controller parameters
kp and kd getting large. As for controller design, the robustness
Fig. 7. Root locus plot. and performance of RPC loop need a trade-off to prevent
instability and keep high bandwidth with low phase delay.
Furthermore, the simulation results need to be verified in the
follow-up experiments.

71

Authorized licensed use limited to: Air University. Downloaded on December 27,2022 at 17:04:24 UTC from IEEE Xplore. Restrictions apply.
REFERENCES Dynamics Simulation,” Journal of Guidance, Control, and Dynamics,
vol. 38, issue 3, pp. 431-441, 2015.
[1] McRuer, D. T., “Aviation Safety and Pilot Control: Understanding and
[11] Vincenzo Muscarello, Giuseppe Quaranta, Pierangelo Masarati, “The
Preventing Unfavourable Pilot-Vehicle Interactions,” Washington, DC:
National Academy Press, 1997. role of rotor coning in helicopter proneness to collective bounce,”
Aerospace Science and Technology, vol. 36, issue 1, pp. 103-113, 2014.
[2] Marilena D. Pavel, Michael Jump, BinhDang Vu, et al, “Adverse
rotorcraft pilot couplings-Past, present and future challenges,” Progress [12] Hess, R. A., “Modeling Human Pilot Adaptation to Flight Control
in Aerospace Sciences, vol. 62, issue 1, pp. 1-51, 2013. Anomalies and Changing Task Demands,” Journal of Guidance, Control,
and Dynamics, Article in Advance, vol. 6, issue 1, pp. 1-12, 2015.
[3] Tustin A, “The nature of the human operator's response in manual
control and its implications for controlled design,” Journal of IEE, vol. [13] Hess, R. A., “Modeling the Pilot Detection of Time-Varying Aircraft
94, issue 1, pp. 190-202, 1947. Dynamics,” Journal of Aircraft, vol. 49, issue 6, pp. 2100-2104, 2012.
[14] Hess, R. A., “Federico Marchesi. Analytical Assessment of Flight
[4] McRuer, D. T., Jex, H. R., “A Review of Quasi-Linear Pilot Models,”
Simulator Fidelity Using Pilot Models,” Journal of Guidance, Control,
IEEE Transactions on Human Factors in Electronics, vol. 8, issue 3, pp.
and Dynamics, vol. 32, issue 3, pp. 760-770, 2009.
231-249, 1967.
[5] McRuer, D. T., Krendel, E. S., Mathematical Models of Human Pilot [15] Hess, R. A., “Simplified Technique for Modeling Piloted Rotorcraft
Behavior. Technical Report, AG-188, AGARD, 1974. Operations Near Ships,” Journal of Guidance, Control, and Dynamics,
vol. 29, issue 6, pp. 1339-1349, 2006.
[6] Hess, R. A., “Theory for Aircraft Handling Qualities Based Upon a
Structural Pilot Model,” Journal of Guidance, Control, and Dynamics, [16] Tischler, M. B., “System identification requirements for high-bandwidth
vol. 12, issue 6, pp. 792-797, 1989. rotorcraft flight control system design,” Journal of Guidance, Control,
and Dynamics, vol. 13, issue 5, pp. 835-841, 1990.
[7] Hess, R. A., “Simplified Approach for Modelling Pilot Pursuit Control
[17] Kemin Zhou, Jone C. Doyle, Keith Glover, “Robust and Optimal
Behaviour in Multi-Loop Flight Control Tasks,” Journal of Aerospace
Engineering, vol. 220, issue 2, pp. 85-102, 2006. Control,” Mao Jianqin, Zhong Yisheng, translated. Beijing, National
Defence Industry Press, pp. 322-326, 2006 (in Chinese).
[8] J. R. Mayo, “The involuntary participation of a human pilot in a
helicopter collective control loop,” 15th European Rotorcraft Forum, [18] US Army AMCOM, “Aeronautical design standard ADS33E-PRF,
Amsterdam, The Netherlands, pp. 1-12, 1989. performance specification, handling qualities requirements for military
rotorcraft,” Redstone, Alabama: ASMC N/A, pp. 76-77, 2000.
[9] Giuseppe Quaranta, Pierangelo Masarati, Joost Venrooij, “Impact of
pilots' biodynamic feedthrough on rotorcraft by robust stability,” Journal
of Sound and Vibration, vol. 332, issue 1, pp. 4948-4962, 2013.
[10] Pierangelo Masarati, Giuseppe Quaranta, Andrea Bernardini et al.,
“Voluntary Pilot Action Through Biodynamics for Helicopter Flight

72

Authorized licensed use limited to: Air University. Downloaded on December 27,2022 at 17:04:24 UTC from IEEE Xplore. Restrictions apply.

You might also like