Aerospace Science and Technology 53 (2016) 241251

Contents lists available at ScienceDirect

Aerospace Science and Technology

www.elsevier.com/locate/aescte

Energy-optimal path planning for Solar-powered UAV with tracking

moving ground target
Yu Huang, Honglun Wang , Peng Yao
School of Automation Science and Electrical Engineering, Beihang University, Beijing 100191, China

a r t i c l e

i n f o

Article history:
Received 14 January 2016
Received in revised form 3 March 2016
Accepted 29 March 2016
Available online 1 April 2016
Keywords:
Solar-powered Unmanned Air Vehicle
(SUAV)
Energy-optimal route
Integrated model
Tracking target
Receding horizon control (RHC)

a b s t r a c t
This paper presents the novel use of receding horizon control (RHC) with particle swarm optimization
(PSO) to generate the energy-optimal trajectories for Solar-powered Unmanned Air Vehicle (SUAV) with
the mission of target tracking. Firstly, an integrated model is presented that accounts for the couplings,
such as kinematics, energetics and mission. The model can formulate the relationship between the
aircrafts position or attitude and the solar time or angle at anytime and anywhere in the world. Next,
the mission of tracking moving ground target is studied in this paper, which is unique relative to past
work about the SUAV. At the same time, to collect more energy and track the moving ground target,
the optimization method of RHC with PSO is used here. To evaluate optimization performance, some
performance indexes are put forward, and the denitions of them are given. Finally, several numerical
simulations demonstrate that this method is feasible and exible to generate the energy-optimal route
for solar-powered UAV online with tracking ground moving target. The analysis of simulations results
indicates that its possible to carry out the task of tracking maneuvering target for a longer time for
SUAV.
2016 Elsevier Masson SAS. All rights reserved.

1. Introduction
In recent years, researchers have increasingly focused on improving the endurance performance for the HAVE/UAV (High Altitude Very-long Endurance/Unmanned Air Vehicle). And Solar-powered UAV (SUAV), e.g. Sky Sailor and Helios, has proved to be an
effective solution with high energy utilization eciency. In Ref. [1],
a solar-powered helicopter prototype is designed and developed. It
was equipped with solar cell on top of its wings, and the consumed energy by the aircraft can be derived from the solar cell
and battery. In [29], the application of optimization techniques
plays a decisive role in increasing the required energy utilization
eciency for SUAV. However, there are still some big challenges.
For example, a suitable mathematical model is dicult to obtain,
which can show the relationship between the attitudes of aircraft
and the solar radiation intensity at any time and any place. In
addition, it is challenging to generate the energy-optimal targettracking trajectories, as there are some complex couplings among
SUAV aerodynamic model, the solar energy harvesting and the mission constraints.
There has been great progress in the recent studies on the solar
powered aircraft. In Ref. [10], the research progress about photo-

Corresponding author.
E-mail address: hl_wang_2002@126.com (H. Wang).

http://dx.doi.org/10.1016/j.ast.2016.03.024
1270-9638/ 2016 Elsevier Masson SAS. All rights reserved.

voltaic cells, rechargeable batteries, maximum power point tracking and so on for aircraft has been reviewed, and some guidance
principles for designers in the design of UAVs are also provided.
A review of the general history and existing literature on the analysis and design of solar-powered vehicles and trajectory planning
is provided in [4,5,8,9,1119]. In [20], implications of longitude
and latitude on the size of solar-powered UAV have been studied.
It is concluded that solar-powered UAVs can be utilized more effectively in the places closer to the equator, where smaller and lighter
solar-powered UAV can be designed. Spangelo et al. [5] put forward
a method to plan a smoother and energy-optimal path in a threedimensional space. In [21], an active power management method
for path planning has been investigated. The usefulness, advantages, and disadvantages of this method over a passive method
are analyzed. To fulll the power requirement with weight constraint of rechargeable batteries, the method of energy stored by
gravitational potential for solar-powered aircraft has been put forward in [22,23]. And the equivalence of gravitational potential and
rechargeable battery for aircraft on energy storage has also been
analyzed.
However, for the denition of energy collection model, the previous studies usually assumed that the solar position and radiation
is invariable and known. Consequently these models may not correctly reect the genuine relationship between solar and SUAV during the long-time ight. In addition, the SUAV must be designed

242

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

Fig. 1. Overall view about sun and aircraft.

based on executing the variety of required missions, and the typical tasks include the intelligence, surveillance and reconnaissance
missions. But in most of literatures such as [3,16,22,23], the optimal path planning of solar aircraft is only based on the simple
mission such as ying from the start position to the nal position.
The energy-optimal path is planned off-line in known static environments, and it may be inapplicable to the dynamic tasks e.g. the
target tacking. To the best of our knowledge, the receding horizon control (RHC) has been used and proved to be an ecient
on-line optimization method in a dynamic environment, which is
based on the simple idea of repetitive solution of an optimal control problem and state updating after the rst input of the optimal
command sequence [24,25].
Based on the above analysis, to achieve the mission of tracking
moving ground target for the SUAV for a long time, an integrated
model is presented and the energy-optimal path planning with
RHC with PSO solver is studied. The simulation results show that
its possible to carry out the task of tracking target for a longer
time for SUAV, with the advantages of simple principle, high computation eciency and good real-time performance. This paper has
the following contributions.
First, by analyzing the relationship between the aircrafts attitude and the solar angle, an integrative model is presented, by
which the radiation on the SUAVs surface can be calculated at anytime and anywhere in the world.
Second, the method of RHC with PSO solver is utilized here,
to plan the on-line energy-optimal trajectory of SUAV for tracking
ground moving target.
Finally, an important conclusion has been drawn that its possible to carry out the task of tracking target for a longer time for
SUAV with the right conditions, e.g. the sucient solar radiation
and the reasonable speed of target.

and the local vertical (down). The computational rules of solar angle are referred in [31], and the denitions of axes and notation
are used the same with Ref. [30].
2.1. Modeling
2.1.1. Aircraft kinematic model
In this paper, the wind axes system is used for SUAV. To simplify the problem, the aircraft is assumed to y in still air at a
constant altitude with the bank-to-turn control scheme, and the
path angle is zero. Hence, aircraft kinematic model is as follows:

dx
dt
dy
dt
d
dt

= V cos

(1)

= V sin

(2)

g tan

(3)

where x is the value of N-axis, y value of E-axis, V is the speed

of aircraft, the yaw angle, the angle of attack, g the constant
of gravity, and is the bank angle.
2.1.2. Solar radiation model
Solar thermal energy can be converted into the electric energy
of UAVs by photovoltaic cell. Solar irradiance, I (W/m2 ), is the rate
at which radiant energy is incident on a unit surface. The suns
radiation is subject to many absorbing, diffusing, and reecting
effects within the earths atmosphere. The extraterrestrial solar radiation perpendicular to the horizontal surface may be calculated
by the following approximate relationship:

2 nday

2

2. Problem formulation

I = I 0 1 + 0.034 cos

In this section, the model and multiple constraints of SUAV are

presented. The integrated model of the aircraft kinematics and energetics, the constraints consisting of maneuvering patterns, and
the airborne electric-optical pod model are formulated, by referring to literatures [3,23,2631].
Fig. 1 shows the overall view, including the composition of
sun, earth, target and aircraft. All their information about position and attitude is expressed in the navigation frame (P-NED in
Fig. 1) which has its origin at the location of the navigation system, point P , and axes aligned with the directions of north, east

where I 0 is the solar constant, and nday is the number of days

(start from January 1 as 1).
In this work, the terrestrial solar radiation is calculated with
ASHRAE Clear Sky Model in [31]. This is a simple statistic model
which is developed based on a large number of simulations using sophisticated spectral simulations and validating with ground
based measurements, which is widely used in the world. This
model is reasonable that the radiation on the wings is calculated,
when the SUAV ies in the stratosphere with almost completely
cloudless sky. The model is described as follows:

365.25

(4)

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

I h = I b sin e + I d

243

(5)

I b = Ie

b mbr

(6)

I d = Ie

d mdr

(7)

b = 1.219 0.043b 0.151d 0.204b d

(8)

d = 0.202 + 0.852b 0.007d 0.357b d

1
mr =
sin e

(9)
(10)

where
I h : solar radiation on a horizontal surface;
I b : beam normal irradiance per unit area normal to the sun rays;
I d : diffuse horizontal irradiance per unit area on a horizontal surface;
mr : air mass ratio;
b , d : beam and diffuse optical depths, the values of which can
be got by looking up table and interpolation;
b, d: beam and diffuse air mass exponents;
e : the elevation angle of the sun.
In practical applications, a rough estimate has been used for
calculating the average value of solar radiation on the surface
earth. Assuming that the radiation is equal in all directions, the
ux of the radiation will be the same. If I T is the total solar radiation output at a distance l from the suns center, the radiation ux
per unit area at a distance l is represented by Q (l), then the total
radiation is equal to 4 l2 Q (l). Hence we can obtain:

IT

Q (l) =
Pe =

4 l 2
Q (l) S e
Sb

Se = r

S b = 4 r

(11)
(12)
(13)

IT
16 l2

a =

(20)

cos() = cos(e ) cos(s a ) sin + sin(e ) cos()

 
2

P s () = I b cos() + I d cos

+ I h r sin

 
2
2

(21)
(22)

where r is the ground reectance factor. The radiation on the

UAVs surface, P uav , is calculated as follows:

(15)

where S e is the sectional area; S b is the area of earths surface; r is

the radius of the earth; P e is the average radiation on the earths
surface (about 340 W/m2 ).
It should be noticed that, all these calculations assume that the
earth is perfectly spherical without any atmosphere and revolves
on a circular orbit without eccentricity.
In this paper, the solar radiation calculation rules are used as
follows. The average radiation on the UAVs surface is subject to
the attitude angles of SUAV, solar azimuth and zenith angle.
According to literature [31], solar azimuth and zenith angle at
any time of a day are described as follows:

sin(e ) = sin(nlat ) sin() + cos(nlat ) cos() cos (t )

cos() cos (t )
sin(s ) =
cos e

We assume that the wing conguration is at and the dihedral

angle is zero. The pitch angle is small when the solar aircraft ies
in still air at a constant altitude in this paper, and the effect of
energy collection power on the UAVs surface due to pitch angle is
negligible. Therefore, the incidence angle of the sun rays upon the
solar cells satises with the expression as follows.
In Fig. 2, ob xb is the shadow of ob xb in the horizontal plane of
earth. ob N  E  is translated from P-NED. a is the tilt angle of aircraft. i is the incident angle of the solar radiation for the wings
surface (angle between sun rays and the normal to the wings surface). We can obtain:

(14)

According to Eqs. (11)(14), we can infer:

Pe =

Fig. 2. Solar angle and attitude of aircraft in the navigation frame.

(16)
(17)

P uav =

P s ()

the accurate calculation

P e cos()

the rough estimate

(23)

2.1.3. Energy harvesting model

The solar cells are mounted on the top side of the wings of
aircraft and gain solar energy from the sun shining on the cells.
The power produced by the photovoltaic panel is:

P in = sol SPuav

(24)

where S is the total surface area of the wing, sol is the eciency
of the solar, and P in is the total power. The value of sol can be
adjusted in accordance with the ratio of full wings surface covered
by solar cells.
In a daytime interval [t 0 , t f ], the energy collected by the aircraft is:

t f

and



= 0.4093 sin 2 (284 + n)/365

(18)

(t ) = 0.2618 (12 t local )

(19)

where s is the azimuth angle of the sun; e represents the elevation angle of the sun; nlat is the latitude; is the declination angle
of sun; (t ) is the hour of sun, and t local is the current hour of the
day.

E in (i ) =

P in dt

(25)

t0

2.1.4. Energy consumption model

Consumption of aircraft mainly contains standard lift, drag and
propulsion models assuming quasi-static equilibrium ight [3].
Consumption of propulsion, P out ( V , ), is dened:

244

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

TV

P out ( V , ) =

prop

T =D
1
D = V 2 SC D
2
CD =
K=

(26)
(27)
(28)

C D 0 + KC 2L

(29)

(30)

R a

Because of aircraft ying in still air at a constant altitude, we

can infer L cos = mg and L = V 2 SC L /2. Hence C L satises:

CL =

2 mg

V 2 S cos

(31)

T is the thrust of the aircraft, prop is the eciency of the propeller, D is the drag of the aircraft, C D is the parasitic drag, K is
the aerodynamic coecient, R a is the aspect ratio of the wing, is
the Oswald eciency factor, L is the lift of the aircraft, and C L is
the coecient of lift.
In a daytime interval [t 0 , t f ], energy consumption of aircraft is

t f
E out (t 0 , t f ) =

P out ( V , )dt

(32)

t0

2.1.5. Airborne electric-optical pod model

The airborne electric-optical pod model is used to localize and
trace target with electric-optical sensor. When the attitudes of aircraft are varying, the view angle of electric-optical sensor remains
unchanged.
Aircrafts position (xb , y b , zb ) and targets position (xc , y c , zc )
in earth coordinate system are shown in Fig. 3. In this work, the
electric-optical sensor is assumed to be installed in the center of
gravity of UAV, and aircrafts position (xb , y b , zb ) is the center of
UAVs gravity. The coordinate point (xo , y o , zo ) is the planar projection of the aircrafts position in earth coordinate system. Then,
the projection (xo , y o , zo ) is satised with equations (33)(35)

xo = xb

(33)

yo = yb

(34)

zo = z c

(35)

And the distance between UAV and moving target is described

as follows:

ltarget =

(xb xc )2 + ( yb y c )2 + ( zb zc )2

(36)

The eld of view is described as follows:

= arccos

H
ltarget

H = zb z c

(37)
(38)

2.2. Multiple constraints

2.2.1. Surveillance task
To achieve the mission of tracking moving ground target, the
distance ltarget and the angle of view should be satised with
the following constraint equations:

ltarget L max

(39)

max

(40)

where L max is the maximum detectable range of airborne electricoptical pod; the value of max represents the maximum angle of
view of electro-optical pod on aircraft.

Fig. 3. Aircraft tracking moving ground target in the navigation frame.

2.2.2. Dynamic constraint

To maintain the stability of vehicle, the two control inputs of
aircraft including the velocity and the bank angle, should satisfy
the following constraints:

Vmin V Vmax

(41)

|| max

(42)

where Vmin is the minimum level ight speed, Vmax the maximum
level ight speed and max the maximum roll angle. In general,
they are determined by the ight vehicle performance. However,
referring to literature [3], Vmin can be set to Vminpower so that aircraft ies at the minimum power.

Vminpower =

4K (mg )2

(43)

3C D 0 2 cos2 ()

3. RHC with PSO for energy-optimal route planning

3.1. PSO
Particle Swarm Optimization (PSO) was rst intended for simulating social behavior, as a stylized representation of the movement
of organisms in a bird ock or sh school. The algorithm has the
advantage of simple principle and it can be observed to be performing optimization [32].
PSO is a metaheuristic method as it makes few or no assumptions about the problem being optimized and can search very large
spaces of candidate solutions. More specically, PSO does not use
the gradient of the problem being optimized, which means it is not
required by PSO that the optimization problem be differentiable
as is required by classic optimization methods such as gradient
descent and quasi-Newton methods. PSO can therefore be used
on optimization problems that are partially irregular, noisy, and
changeable over time, etc.
It is demonstrated that PSO can nd better results in a faster,
cheaper way compared with other methods [24]. In this work, PSO
is used for searching optimum value in given space at every step.
The PSO algorithm is described as

v j +1 (k, n) = w v j (k, n) + c 1 r1 (0, 1) p pest (k, n) x j (k, n)

j

 j

+ c 2 r2 (0, 1) g pest (k, n) x j (k, n)

j +1

(k, n) = x (k, n) + v (k, n)

k = 1, 2, . . . , N p
n = 1, 2, . . . , D num

(44)
(45)

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

Subject to

a j x j (k, n) b j

(46)

v (k, n)
|b j a j |

(47)

D num = N var N

(48)

where N p represents the number of particle population, the value

of which should be chosen by considering complexities and diversities of optimization problem, and computational time at same
time. The bigger N p is, the longer time PSO costs. D num represents the search space dimension. N var represents the number of
control variables in RHC about aircraft kinematic model. N is the
length of future control sequence in RHC. r1 (0, 1) and r2 (0, 1) are
two independent random numbers within the range [0, 1]; c 1 and
c 2 are two learning factors; a j and b j are the upper bound and
lower bound of particle position. In this paper, each particle represents a possible solution to the constrained optimization problem
and results in a specic value of the objective function.
3.2. RHC
The RHC method based on PSO is adopted in this paper. RHC
is an optimizing control algorithm within limited time domain,
where the future state of the controlled plant could be predicted
using the prediction model and optimized in receding horizons [6].
It has been proved to be more successfully optimized online in a
dynamic environment, which is based on the simple idea of repetitive solution of an optimal control problem and state updating
after the rst input of the optimal command sequence [24]. The
main idea of RHC is the online receding/moving optimization. It
breaks the global control problem into several local optimization
problems of smaller sizes, which can signicantly decrease the
computing complexity and computational expense. Based on the
updated information, the UAV will optimize the local paths with
the regular horizon by PSO. Then the results are executed. The
above steps repeat until the mission ends.
The process of RHC method is as follows. At time k, suppose
that u [k : k + N 1] = [u (k), , u (k + N 1)] is the future control
sequence of length N, x(k) is the state vector, and J (x(k), u [k :
k + N 1]) is the objective function by evaluating the effects of
the future control sequence. The optimal control sequence u [k :
k + N 1] is obtained by minimizing the objective function:

min J = J x(k), u [k : k + N 1]
u

(49)

Given the accuracy of the prediction, only the rst term i.e.
u (k) is executed, and the other terms i.e. u [k + 1 : k + N 1]
are taken as the initial guess of control inputs for the next step.
During the execution, the new control sequence for the next step
is planned. And the planning time should be less than the execution time.
For the denition of total objective function, the control objective function J c and the reconnaissance object function J r are
considered at the same time:

J x(k), u [k : k + N 1] = J c + J r

(50)

3.2.1. Control objective function

The control object is to obtain the total energy as much as possible. However, considering Eq. (49), the value of the total energy
must be multiplied by 1 as follows

J c = 1

k+
N 1




t P(i )in P(i )out

(51)

i =k+1

where, 1 is the weight coecient, the value of which will be set

depending on the actual problem. t the time step; P(i )in and

245

P(i )out are the aircrafts energy harvesting power and energy consumption power, which are dened by Eq. (24) and Eq. (26).
3.2.2. Reconnaissance objective function
The reconnaissance objective function will be chosen according
to requirement of reconnaissance task. The reconnaissance objective function is formulated as follows:


Jr =

2
2

k+ N 1
i =k+1

k+ N 1
i =k+1

if i max

li
li

( L max )2li otherwise

(52)

where li is the distance between UAV and moving target at the

time i, and it is dened by Eq. (36), i is the angle of view at the
time i, 2 is the weight coecient of object function.
3.3. RHC with PSO implementation procedure
The detailed implementation procedure of RHC with PSO for
Energy-Optimal path planning can be described as follows.
Step 1: According to the environmental and the optimization
problem models in Section 2, initialize the beginning, nal information and reconnaissance information.
Step 2: Initialize parameters of PSO algorithm, such as search
space dimension D, the population size N p and the maximum
number of the iterations N it , and set the beginning values of x(k)
and v (k) about each particle.
Step 3: Initialize parameters of RHC algorithm, such as the
length of future control sequence N, the objective function.
Step 4: Initialize the position and velocity of each particle, and
compare the tness of each particle to nd the current best control
command.
Step 5: Update particle velocity and position using Eqs. (44)
(45), and choose the optimal control sequences at the current iteration j by comparing the tness of each group particle. Repeat
this step until j reaches the maximum iteration.
Step 6: Calculate the optimal trajectory with the rst term of
the optimal control sequences using Eqs. (1)(3).
Step 7: If the mission does not end, go to step 4.
The above steps can be summarized as pseudo code (see Table 1).
4. Simulation and results
In this section, the parameters of aircraft and battery pack are
listed in Table 2 from [23,27]. At rst, the simulation results about
energy harvesting model with specic attitude of UAVs have been
given in Beijing (39.93 N, 116.28 E) throughout the year. Then,
we present the simulation results of energy-optimal path planning with tracking moving ground object on January 21 using the
method of RHC with PSO. More ever, the total collection energy
and the tracking success ratio are introduced to evaluate the optimization results. (See Table 2.)
4.1. Numerical simulation of energy collection model
According to Eqs. (4)(24), the simulations of instantaneous
power produced by solar cell form solar radiation on the SUAVs
wing on Jan 21, Apr 21, Jul 21 and Oct 21 in Beijing are shown in
Fig. 4, which are chosen in different seasons.
As shown in Fig. 4, the instantaneous power at any time on
Jul 21 in the summer season is the biggest compared with that of
other days solar radiation, and in general, the power reaches the
maximum value at noon in every day.

246

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

Table 1
Proposed algorithm.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

//Initialization
Set the parameters of the aircraft model: prop , , S , C D 0 , g, m, , K , AR
Set the parameters of the solar radiation model: I 0 , b , d
Set the parameters of the algorithm: N var , N , N p , N it , c 1 , c 2 , w
Set the initial sates of the vehicle: nlat , nlon , nday , t local , V 0 , 0 , 0 , zb0
Set the initial sates of the objection: xc0 , y c0 , zc0
Set the end condition: T f
//Main loop
While t local T f (not reaching the stop criteria) do
Generate the solar radiation using Eqs. (4)(10), Eqs. (16)(23)
for N it iterations
Generate the population with N p random particles with constraint Eqs. (41), (42)
Update particle velocity and position using Eqs. (44), (45)
Calculate the objection function using Eqs. (49)(52)
Determine the personal best position and global best position
end for
Generate the optimal control sequences according to global best position
Calculate the optimal trajectory with the rst term of the optimal control sequences using Eqs. (1)(3)
Update the initial sates of UAV with the optimal trajectory
t local = t local + t
end while

Fig. 4. The instantaneous power on different day throughout the year.

4.2. Energy-optimal path planning

Table 2
Aircraft model parameters [23,27].

In this section, the simulations consist of three parts: a brief

comparison of path planning with/without energy optimization,
the inuence of main parameters on the optimization method and
the optimization results with different target velocities. To mathematically analyze and adjust optimal design, some performance
indexes will be introduced. In this paper, the total energy of SUAV
at the end of ight and the tracking success ratio are dened.
The total energy, E total , is the difference between the collected
energy and energy consumption. The optimization strategy will be
better if the total energy is bigger. It can be described as

E total = E in () E out (t 0 , t f )

(53)

The tracking success ratio, , is dened as the time meeting the

requirements of mission to the total time ratio. The optimization
strategy will be better if the tracking success ratio is bigger. It can
be described as

ntask t
t f t0

(54)

where, ntask is the number of waypoints meeting the relation of

Eqs. (39), (40), t is the time interval in two adjacent waypoints.
In these simulations, the equation of motion of the moving
ground target is described as

Parameters

Notation

Value

Unit

Wing area
Mass
Wingspan
Oswald eciency factor
Parasitic drag
Propeller eciency
Eciency of the solar cell
Air density
Battery capacity

S
m
b

0.1566
1.2
0.711
0.992
0.011
0.7
0.22
1.29
30

m2
kg
m

CD0

prop
sol



A
x = t + B 2 rank(0, 1) 1
2


A
y = t + B 2 rank(0, 1) 1
2

kg/m3
W

(55)
(56)

where, x and y are components of N axis and E axis respectively

in the navigation frame (P-NED) whose origin is set in Beijing. A
represents the speed of ground object. B (2 rank(0, 1) 1) represents the error of target position.
4.2.1. Path planning with/without solar energy optimization
In this case, a brief comparison of path planning with/without
solar energy optimization is presented. The simulation parameters
are set by Table 3. To make SUAV path planning, we have an intu-

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

247

Table 3
Simulation parameters.
Place: Beijing

Date: January 21

Begin time: 12:30 pm

End time: 12:35 pm

Parameters

Notation

No optimized

Optimized

The number of particle population

The number of the iterations
The length of future control sequence
The number of control variables
The weight coecient of control object function
The weight coecient of reconnaissance objective function
The ight altitude
The maximum roll angle
The speed of UAV with the minimum power
The maximum level ight speed
The time step
The maximum detectable range
The maximum angle of view
The objects speed
The error of objects position

Np
N it
N
N var

60
20
5
2
0
1
300 m
20
15 m/s
30 m/s
1s
360 m
70
5
5

60
20
5
2
5
1
300 m
20
15 m/s
30 m/s
1s
360 m
70
5
5

1
2
z

max
V minpower
V max
t
L max

max
A
B

Fig. 5. Tracking moving ground object without energy optimization.

Fig. 6. Tracking moving ground object with energy optimization.

Fig. 7. The energy harvesting curve with time.

248

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

Fig. 8. The tracking success ratio curve with time.

Fig. 9. Tracking path with the different value of N.

Fig. 10. The energy harvesting curve with different N.

itive feeling for energy collection, and the instantaneous power is

dened as follows:

P i = P in P out ( V , )

(57)

If P in P out ( V , ), the waypoint of SUAV is plotted in simulation gure with the form of plus. If P in < P out ( V , ), the waypoint
of SUAV is plotted in simulation gure with the form of the lled
circle. Where, P i represents the total UAVs power at ith time. If
P i is negative, the total collection energy stored in rechargeable
batteries on the solar airplane will decrease.

Fig. 5 and Fig. 6 illustrate the trajectories of moving target and

solar aircraft. Its very easy to see that the sum of positive instantaneous power in Fig. 6 is more than the other. From Fig. 7, the
total energy with energy optimization is more than doubled the
amount without energy optimization. However, their tracking success ratios are very close, as shown in Fig. 8.
4.2.2. The effect of main parameters on the optimization design
In this section, we mainly analyze the length of future control
sequence N that represents the xed future time interval and the

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

249

Fig. 11. The tracking success ratio curve with different N.

Fig. 12. Tracking path with different detectable range.

Fig. 13. The energy harvesting curve with different detectable range.

maximum detectable range. In the following simulations, the simulation parameters are the same with Table 3 with the exception
of N and L max .
The optimized paths with different N are given in Fig. 9, and
their corresponding performances are shown in Fig. 10 and Fig. 11.
By analyzing Fig. 10 and Fig. 11, N = 3 is a reasonable choice, comparing with other values. In a general way, the value of N can

be decided by cut and try methods for practical matters, which is

nearly optimum.
Figs. 1214 are the simulation results by RHC with PSO with
different detectable ranges. In this paper, we assume that the angle
of view is big enough so that it has no inuence upon the total
collection energy and tracking success ratio. By analyzing Fig. 13
and Fig. 14, the total energy and the tracking success ratio will be
better if the value of maximum detectable range is bigger.

250

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

Fig. 14. The tracking success ratio curve with different detectable range.

Fig. 15. Tracking path with different object velocities.

Fig. 16. The energy harvesting curve with different object velocities.

4.2.3. The optimization results with different target velocities

In this case, the simulation parameters are the same with Table 3 except the target speed. By analyzing Figs. 1517, it shows
that the aircraft has to do large maneuver to keep the solar aircraft
with preferable tracking success ratio and satised total energy, if
the moving objects velocity is low. If the target moves too fast
(e.g. 21 m/s), the total energy will be negative, which shows that
the task executed for a long time is not feasible. And more, it is
easy to understand that the smaller the speed of aircraft is, the

less the energy consumption is, and the more the total energy will
be in the given time interval with the same collection energy.
5. Conclusions
The design of energy-optimal trajectory contributes to increasing endurance for the solar aircraft. By the presented method, it is
possible to carry out the task of tracking target for a long time for
SUAV.

Y. Huang et al. / Aerospace Science and Technology 53 (2016) 241251

251

Fig. 17. The tracking success ratio with different object velocities.

1) In this paper, we present a simple energy integrated model in

two-dimensional space to calculate the instantaneous power
at any time in any place.
2) The generation of waypoint for solar aircraft to track moving
ground target is carried out by using the constrained RHC with
PSO algorithm.
3) Based on the energy integrated model and the control design,
the effect of main parameters (the size of the sliding window,
the detectable range and the target velocity) on optimization
results are discussed.
Future work will focus on the research about energy-optimal
path planning in three-dimensional space with uncertainty in complex environment.
Conict of interest statement
The authors declared that they do not have any conicts of interest to this work.
Acknowledgements
The authors would like to thank the editors and reviews for
their critical review of the manuscript. This research has been supported by the Changjiang Scholars and Innovative Research Team
in University under Grant IRT 13004, in part by Aeronautical Science Foundation of China under Grant 2014ZA51002, and in part
by the National Natural Science Foundation of China #61175084.
Appendix A. Supplementary material
Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.ast.2016.03.024.
References
[1] M.H. Shaheed, et al., Flying by the Sun only: the Solarcopter prototype, Aerosp.
Sci. Technol. 45 (2015) 209214.
[2] S. Hosseini, R. Dai, M. Mesbahi, Optimal path planning and power allocation for
a long endurance solar-powered UAV, in: 2013 American Control Conference
(ACC), 2013, pp. 25882593.
[3] A.T. Klesh, P.T. Kabamba, Solar-powered aircraft: energy-optimal path planning
and perpetual endurance, J. Guid. Control Dyn. 32 (4) (2009) 13201329.
[4] J.K. Shiau, et al., Design of a solar power management system for an experimental UAV, IEEE Trans. Aerosp. Electron. Syst. 45 (4) (2009) 13501360.
[5] S.C. Spangelo, E.G. Gilbert, Power optimization of solar-powered aircraft with
specied closed ground tracks, J. Aircr. 50 (1) (2013) 232238.
[6] X.-M. Tang, P. Chen, B. Li, Optimal air route ight conict resolution based on
receding horizon control, Aerosp. Sci. Technol. (2015).
[7] W. Harasani, et al., Initial conceptual design and wing aerodynamic analysis of
a solar power-based UAV, Aeronaut. J. 118 (1203) (2014) 540554.

[8] X.Z. Gao, et al., Parameters determination for concept design of solar-powered,
high-altitude long-endurance UAV, Aircr. Eng. Aerosp. Technol. 85 (4) (2013)
293303.
[9] E. Cestino, Design of solar high altitude long endurance aircraft for multi payload & operations, Aerosp. Sci. Technol. 10 (6) (2006) 541550.
[10] X.-Z. Gao, et al., Reviews of methods to extract and store energy for solarpowered aircraft, Renew. Sustain. Energy Rev. 44 (2015) 96108.
[11] P. Panagiotou, et al., Aerodynamic design of a MALE UAV, Aerosp. Sci. Technol.
(2016).
[12] J.K. Shiau, D.M. Ma, Development of an experimental solar-powered unmanned
aerial vehicle, J. Chin. Inst. Eng. 38 (6) (2015) 701713.
[13] A. Malaver, et al., Development and integration of a solar powered unmanned
aerial vehicle and a wireless sensor network to monitor Greenhouse gases, Sensors 15 (2) (2015) 40724096.
[14] E. Troncoso, N. Lapena-Rey, O. Valero, Off-grid test results of a solar-powered
hydrogen refuelling station for fuel cell powered Unmanned Aerial Vehicles,
Int. J. Hydrog. Energy 39 (21) (2014) 1126711278.
[15] J.K. Shiau, et al., Optimal sizing and cruise speed determination for a solarpowered airplane, J. Aircr. 47 (2) (2010) 622629.
[16] J. Meyer, et al., Design considerations for a low altitude long endurance solar
powered unmanned aerial vehicle, in: 8th IEEE Africon Conference, September
2007, IEEE, 2007, pp. 507512.
[17] B. Lee, et al., The ight test and power simulations of an UAV powered by solar
cells, a fuel cell and batteries, J. Mech. Sci. Technol. 28 (1) (2014) 399405.
[18] X.F. Zhu, et al., How high can solar-powered airplanes y, J. Aircr. 51 (5) (2014)
16531659.
[19] G. Marques Hobold, R.K. Agarwal, A methodology for predicting solar power
incidence on airfoils and their optimization for solar-powered airplanes, Proc.
Inst. Mech. Eng., Part G, J. Aerosp. Eng. 229 (7) (2014) 12671279.
[20] P. Rajendran, H. Smith, Implications of longitude and latitude on the size of
solar-powered UAV, Energy Convers. Manag. 98 (2015) 107114.
[21] B. Lee, et al., Active power management system for an unmanned aerial vehicle
powered by solar cells, a fuel cell, and batteries, IEEE Trans. Aerosp. Electron.
Syst. 50 (4) (2014) 31673177.
[22] X.Z. Gao, et al., The equivalence of gravitational potential and rechargeable battery for high-altitude long-endurance solar-powered aircraft on energy storage,
Energy Convers. Manag. 76 (2013) 986995.
[23] A. Klesh, P. Kabamba, Energy-optimal path planning for solar-powered aircraft
in level ight, in: AIAA Guidance, Navigation and Control Conference and Exhibit, 2007.
[24] H. Duan, Y. Zhang, S. Liu, Multiple UAVs/UGVs heterogeneous coordinated technique based on Receding Horizon Control (RHC) and velocity vector control,
Sci. China, Technol. Sci. 54 (4) (2011) 869876.
[25] P. Yao, H. Wang, Z. Su, Real-time path planning of unmanned aerial vehicle
for target tracking and obstacle avoidance in complex dynamic environment,
Aerosp. Sci. Technol. 47 (2015) 269279.
[26] D.M. Ma, et al., Optimal level turn of solar-powered unmanned aerial vehicle
ying in atmosphere, J. Guid. Control Dyn. 33 (5) (2010) 13471356.
[27] S. Hosseinisianaki, M. Mesbahi, Energy aware aerial surveillance for a long endurance solar-powered UAV, in: AIAA Guidance, Navigation, and Control, GNC,
Conference, 2013.
[28] Q. Dai, X. Fang, A simple model to predict solar radiation under clear sky conditions, Adv. Space Res. 53 (8) (2014) 12391245.
[29] S. Pattarapanitchai, S. Janjai, A semi-empirical model for estimating diffuse solar irradiance under a clear sky condition for a tropical environment, Proc. Eng.
32 (2012) 421426.
[30] D. McLean, Automatic Flight Control Systems, Prentice Hall International (UK)
Ltd., 1990.
[31] D.Y. Goswami, Principles of Solar Engineering, third edition, CRC Press, 2015.
[32] J. Zhao, R. Zhou, Particle swarm optimization applied to hypersonic reentry
trajectories, Chin. J. Aeronaut. 28 (3) (2015) 822831.