Вы находитесь на странице: 1из 9

JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.

1 (1-9)
Aerospace Science and Technology ()

1 67
Contents lists available at ScienceDirect
2 68
3 69
4 Aerospace Science and Technology 70
5 71
6 72
7 www.elsevier.com/locate/aescte 73
8 74
9 75
10 76
11 77
12 Entry trajectory generation without reversal of bank angle 78
13 79
14 Ruizhi He a, , Luhua Liu a , Guojian Tang a , Weimin Bao a,b 80
15 81
a
College of Aerospace Science and Engineering, National University of Defense Technology, Changsha 410073, China
16 b 82
China aerospace science and technology corporation, Beijing 100048, China
17 83
18 84
19 a r t i c l e i n f o a b s t r a c t 85
20 86
21 Article history: A rapid entry trajectory planning algorithm for hypersonic glide vehicles is presented in this paper, and 87
22 Received 19 October 2016 the avoidance of no-y zones with large cross-range is achieved. The highly-constrained entry trajectory 88
23 Received in revised form 10 October 2017 generation mission is accomplished using an adjustable angle of attack and bank, instead of a nominal 89
Accepted 10 October 2017
24 angle of attack and reversal logic of bank angle. In this way, the maneuvering capability of the vehicle is 90
Available online xxxx
25 greatly improved. Moreover, as no reversal of bank angle exists in the whole ight, the control command 91
26 Keywords: is easier to follow. Firstly, longitudinal entry corridor with an adjustable angle of attack is built up. 92
27 Entry trajectory The drag acceleration prole is represented as ve piecewise ane functions of the energy, which are 93
Rapid generation adjusted to meet the requirements of the trajectory length and lateral maneuvering capability. Secondly,
28 94
No-y zone based on the quasi-equilibrium glide condition (QEGC), the maximum and minimum boundary of the
29 Adjustable angle of attack 95
lateral lift-to-drag corridor corresponding to the current drag acceleration prole is approximated, and
30 No reversal of bank angle 96
the lateral prole is represented as an interpolation result of the boundaries. Finally, the parameters of
31 longitudinal and lateral prole are found by the Newton iteration scheme with the help of a reduced 97
32 order system. A combined proportional integral derivative (PID) tracker is also designed to follow the 98
33 drag and heading angle prole. The approach is tested using the Common Aero Vehicle model. Simulation 99
34 results demonstrate that the generated entry trajectories can avoid no-y zone with a large cross-range 100
35 and achieve the terminal conditions while satisfying all the complex path constraints. 101
36 2017 Elsevier Masson SAS. All rights reserved. 102
37 103
38 104
39 105
40 106
41
1. Introduction (SOCS), the entry trajectory with some common path constraints 107
42
is obtained. However, it needs large amount of computing power. 108
43 Nominal angle of attack and lateral azimuth error deadband are To solve this problem, Zhang [12] proposed a three-dimensional 109
44 widely used in the entry trajectory generation. The Shuttle entry acceleration prole (TDAP) planning method, and generated the 110
45 guidance [1], i.e. drag based guidance, assumes that the entry tra- entry trajectory in a limited corridor. Furthermore, trajectory gen- 111
46 jectory is a great circle, and the angle of attack is designed as a eration methods without nominal angle of attack prole were car- 112
47 linear function of velocity or normalized non-conventional energy. ried out. For example, Saraf [13] proposed an angle of attack based 113
48 The magnitude of the bank angle is adjusted to follow the ref- controller to track the heat rate curve in the altitude versus veloc- 114
49 erence drag prole. Several reversals of bank angle in the lateral ity plane, and make it possible to compute the enlarged landing 115
50 azimuth error corridor are used to minimize the cross-range error footprint. By sequentially optimizing the weighting matrices for 116
51 at the terminal area energy management (TAEM) interface. Shut- bank-angle control and angle of sideslip control, Mooij [14] halved 117
52 tle entry guidance has been proven effective for the current space the number of design parameters, and reduced the non-linearity 118
53 transportation system. Recently, some further improvements have of the design space. On the other hand, Shen [15] proposed a 119
54 been made, such as the simplication of the reference drag pro- guidance logic, which determines the bank reversals by constantly 120
55 le [2,3], the new shape of the error corridor [4] and the accuracy evaluating information from the reference cross-range prole, cur- 121
56 enhancement of the tracker [59]. However, these efforts are still rent cross-range, and estimated actual lift-to-drag ratio. Luo [16] 122
57 within the connes of the basic shuttle entry guidance method. presented a novel lateral guidance algorithm based on analytical 123
58 A 3-DOF prole planning algorithm was rst proposed by Mease derivations and formed a patched corridor for low-lifting vehi- 124
59 [10] and Saraf [11]. With the Sparse Optimal Control Software cles. 125
60 As the further application of QEGC to the trajectory generation, 126
61 some simplied dynamic models and parametric methods are de- 127
62 * Corresponding author. veloped. For example, Wang [17] developed a predictorcorrector 128
63 E-mail address: heruizhi_nudt@sina.com (R. He). guidance law based on fuzzy logic. Xu [18] proposed an entry 129
64 130
https://doi.org/10.1016/j.ast.2017.10.019
65 131
1270-9638/ 2017 Elsevier Masson SAS. All rights reserved.
66 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.2 (1-9)
2 R. He et al. / Aerospace Science and Technology ()

1 guidance method based on a reduced-order system and a paramet- In this study, a rapid entry trajectory planner is proposed. En- 67
2 ric bank angle. Since the analytical predictive of trajectory length try trajectories with large cross-range satisfying no-y zone and 68
3 and terminal velocity, the parameters are corrected according to other path and terminal constraints are achieved. The approach 69
4 the desired terminal condition. In order to reduce the calculation is an extension of the drag acceleration guidance. Both the longi- 70
5 time and make an onboard generation, Sagliano [19] proposed an tude drag prole and lateral lift to drag prole are generated with 71
6 onboard trajectory generation for entry vehicles via adaptive mul- an adjustable angle of attack and bank. No reversal of bank angle 72
7 tivariate pseudospectral interpolation, and the oine and online is operated in the whole ight. No-y zone is avoided effectively, 73
8 planning are combined in this method. Based on the bank reversal and both the downrange and cross-range are satised by utilizing 74
9 logic and several simplied models, Yu [20] developed autonomous 75
both the angle of attack and bank angle. This paper is organized
10 entry guidance for hypersonic glide vehicle using 3-D analytical 76
as follows. Section 2 describes the trajectory generation problem,
11 glide formulas. Shen [21] proposed a parameter iterative algorithm 77
including entry dynamics and constraints. Section 3 establishes
12 78
to search for the magnitude and reversal point of bank angle with an entry planning algorithm, which generates both the longitude
13 79
a nominal angle of attack, and the entry trajectory could be gener- drag prole and lateral lift to drag prole, considering complex
14 80
ated in several seconds on a desktop computer. Similarly, using a path constraints. Also, a PID tracker is proposed to follow both the
15 81
maneuver coecient, Liang [22] proposed a parametric trajectory drag prole and heading angle prole. Section 4 demonstrates the
16 82
generation method, in which the designs for the longitudinal and validity of planning algorithm with the PID tracker using the Com-
17 83
lateral trajectories are decoupled. Furthermore, Lu [23] developed mon Aero Vehicle model, and examines the adaptive ability of the
18 84
an entry trajectory planning algorithm by solving an optimization proposed approach under different terminal positions. Simulation
19 85
problem, and an analytical bank angle has been achieved with results of traditional method and the proposed method are also
20 86
different optimal indexes. However, all these methods mentioned compared in this section. Finally, Section 5 summarizes the main
21 87
above set the design of bank angle separated from the angle of contributions and the further works for future research.
22 88
attack. Moreover, multiple parametric models and lateral reversal
23 89
logic are used, which greatly limits the maneuvering capability.
24 2. Trajectory generation problem 90
Thus, a series of methods making the combined use of angle of
25 91
attack and bank angle were proposed. Mooij [24] developed an
26 92
adaptive heat-ux tracking algorithm by a linear output-feedback 2.1. Entry dynamics
27 93
controller and a guidance-tracking system, based on so-called sim-
28 94
ple adaptive control. Bollino [25,26] investigate the utility of a The hypersonic glide vehicle is unpowered, and there exists a
29 95
pseudospectral guidance algorithm for solving onboard trajectory non-conservative force (aerodynamic force) doing negative work
30 96
generation problem, and a revolutionary concept based on the during the whole entry ight. The energy E is a monotonically de-
31 97
principles of nonlinear optimal control theory was set up. Similarly, creasing variable which can be expressed as
32 98
33
Sagliano [27] proposed a trajectory generation algorithm based on 99
high-order sliding-mode theory, and described in detail two differ-
E = v 2 /2 /r
34 100
ent guidance systems [28] developed for SHEFEX-3, main guidance (1)
35 101
36
and backup guidance, which can potentially be used for the SHE- 102
FEX program evolution. Also, Lu [29] proposed a robust trajectory where v is the magnitude of the velocity with respect to the Earth,
37 103
generation algorithm, in which the nonlinear trajectory control law and can be solved as
38 104
39 is derived by continuous minimization of the predicted tracking er-  105
40 ror. v= 2( E + /r ) (2) 106
41 For some special entry missions [30], no-y zone located along 107
42 the ight-path must be avoided. The trajectory generation is trans- is the gravitational constant, and r is the radius from the vehi- 108
43 formed into a very dicult trajectory problem with complex path cle to the Earth center. In order to facilitate the description of the 109
44 and terminal constraints. Zhang [31] and Xie [32] put forward a motion, E is treated as the variable of integration [32]. 110
45 lateral reversal logic combined with the longitude drag prole gen- 111
46 eration, by which entry vehicles could get effective avoidance of dr sin 112
47 the no-y zone according to the center position and threat radius. = (3) 113
dE D
48 Liang [33,34] proposed a bank reversal strategy based on the tra- 114
jectory prediction, and the trajectory generation under location and d cos sin
49
= (4) 115
50 direction constraints is solved. Similarly, in [35], the no-y zone dE r D cos 116
51 was transformed into the edge point constraint, and several re- 117
d cos cos
52 versals were assumed in the parametric model of bank angle. The = (5) 118
53 trajectory is generated in segments while satisfying all the path dE rD 119
and terminal constraints. Jorris [36] and Zhang [37], proposed a  
54 d L v2 cos 120
55 direct trajectory optimization approach, in which the control and = cos + g + C1 + C2 (6) 121
dE v2 D r v2 D
56 state variables were discretized in subtle ways. The ight time 122
57 was minimized and the avoidance of no-y zone constraint was d tan cos sin L sin 123
58 achieved. In recent years, entry trajectory planning algorithms con- = + C 1 + C 2 (7) 124
dE rD v2 D cos
59 sidering no-y zone constraint has gained great attention in recent 125
60 years. Reversal logic of bank angle is widely used and the bank an- where is the longitude, is the latitude, is the ight-path 126
61 gle is usually utilized as the only primary control command. Angle angle and is the heading angle measured from the north in 127
62 of attack is slightly adjusted during the reversal, which ensures a clockwise direction. The gravitational acceleration is g = /r 2 . The 128
63 smooth variation of the altitude. The only 3-D proles generation control variables used are the angle of attack and bank angle 129
64 method uses the optimal software, which needs large amount of , where the positive direction of in our notation is clockwise. 130
65 computing and consumes time and can not meet the requirement C 1 , C 2 , C 1 , C 2 are several variables relating to the angular rate 131
66 of next generation vehicles. of planet rotation e , which are expressed as [32]: 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.3 (1-9)
R. He et al. / Aerospace Science and Technology () 3


2e
1
C1 = sin cos 67

VD
2

68

2
r
3

C 2 = 2e (cos sin cos sin + cos2 cos )
69
4 70
V D (8)
5
2e 71

C 1 = (sin cos tan cos )
6

VD
72


7

2 r sin cos sin 73
8 C 2 = e 74
9 V 2 D cos 75
Fig. 1. Ground track under no-y zone constraint.
10 The lift and drag accelerations are given by 76
11 77
1
In glide phase, the algorithm transfers the highly constrained
12 L= (r ) v 2 S r C L , M ( E , r ) /m 78
13 2 trajectory planning problem into longitudinal and lateral prole 79
(9)
1 2

generation problems. Adjustable and continuous angle of attack
14
D= (r ) v S r C D , M ( E , r ) /m and bank are used in this method, instead of the nominal angle
80
15 2 81
of attack and reversal logic of bank angle. No-y zone constraint
16 The atmospheric density is given as the form of (r ) = 82
is also considered in the trajectory generation process. Highly con-
17 0 exp[(r R 0 )/h s ], where 0 = 1.225 kg/m3 , R 0 is the Earth 83
18 strained entry trajectory generation mission is accomplished with 84
radius, and h s is the scale height with the value of 7110 m. The
a greatly improved maneuvering capability. As no reversal of bank
19 reference area S r (m2 ) and mass m (kg) are two typical character- 85
20 angle exists in the whole ight, the diculty of the tracking sys- 86
istic parameters of the entry vehicle, and C L , C D are the lift and
21 tem is also reduced effectively. Fig. 2 shows the owchart of the 87
drag coecients.
22 glide trajectory generation algorithm. 88
23 The steps for the planning algorithm are shown as follows: 89
2.2. Vehicle and control constraints
24 Step 1: Set up a new DE ight corridor with an adjustable 90
25 angle of attack, based on the range of the angle and several path 91
Heating rate Q , dynamic pressure q, and total load n can be
26 constraints; 92
expressed as [35]:
27 Step 2: Employ the longitudinal sub-planner to design the drag 93
28 Q = k 0.5 v 3.15 Q max acceleration prole based on the desired maneuvering capability 94
29 and trajectory length; 95
2
q = v /2 qmax (10) Step 3: Set up the corresponding lateral corridor to the drag
30
 acceleration prole employed in step 2;
96
31 n= L 2 + D 2 / g 0 nmax 97
32 Step 4: The lateral prole is designed as the interpolation re- 98
33 where k is a constant with the value of 5e 5 (kw s3.15 /kg0.5 sult of the upper and lower boundaries of lateral corridor. Employ 99
34 m3.65 ), which is related to the shape of the vehicle, g 0 is gravita- the lateral sub-planner to calculate the interpolation coecient 100
35 tional acceleration at R 0 . Further, the QEGC condition is considered according to the desired cross-range. Newton iteration scheme is 101
36 as a soft constraint to ensure the atness of the ight trajectory used in this method to calculate D 1 , D 2 , from the beginning of 102
37 D 1 = 2 m/s2 , D 2 = 3 m/s2 , = 0.5, and minimize the terminal 103
38 L cos + v 2 /r g = 0 (11) location error; 104
39 Step 5: If both the crossrange and the trajectory length are sat- 105
Suppose the entry ight ends at the terminal area energy man- ised, go to step 6; otherwise, repeat Steps 25;
40 106
agement interface, the terminal altitude h f , terminal velocity v f Step 6: Calculate the control variables re f , re f based on the
41 107
and the range to the target S f should be: generated drag acceleration D re f and lateral lift to drag prole
42 108
43 h f = hT A E M , v f = v T A E M , S f = S T A E M (12) ( L sin / D )re f . As the reference heading angle is also gained in the 109
44 lateral sub-planner, the PID tracker is used to follow both the drag 110
Control constraint is also an important part of the path con- prole and heading angle prole. Thus, all the control variables
45 111
straints: and trajectory states are gained effectively.
46 112
47 [min , max ], [min , max ] 113
(13) 3.1. Longitudinal sub-planner
48 | | max , | | max 114
49 115
Typical path constraints like heating rate, dynamic pressure,
50 3. Entry planning algorithm 116
51
total load and QEGC condition can be transformed into the con- 117
straints of drag acceleration[35]:
52 In initial phase, as the atmospheric density is too thin, simple 118
53 control form [32] is used with a constant angle of attack and bank 2 119
C D S r Q max
54 until the vehicle is steered into the QEGC condition. In glide phase, D ( E ) D Q ( E ) = 120
2mk v 4.3
2
55 adjustable angle of attack and bank are assumed. No-y zone is nmax g 0 (14) 121
56 successfully avoided and complex missions are accomplished while D (E ) Dn (E ) =  122
57 satisfying other path constraints. In this paper, glide trajectory gen- 1 + ( L / D )2 123
58 eration is the key point. qmax C D S r 124
D(E ) Dq(E ) =
59 Fig. 1 shows the avoidance process from the upper side of the m 125
60 no-y zone. Point A is the transition point between initial phase   (15) 126
D v2
61 and glide phase. Point B is the test point on the edge of the no-y D ( E ) D eg ( E ) = g 127
L r
62 zone, which guarantees the entry trajectory is far enough from the 128
63 center. Point T is the desired target. The task of the glide trajectory Considering that the angle of attack is adjustable, it is dicult 129
64 generation is to plan the trajectory AB and BT, steer the vehicle to set up the new DE ight corridor. Since the altitude variation 130
65 into the desired target with a preselected energy, and satisfy no-y of h is small compared to the magnitude of r in glide phase, we set 131
66 zone and other path constraints. r = (r0 + r f )/2. Thus, at each value of energy E, all the inequality 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.4 (1-9)
4 R. He et al. / Aerospace Science and Technology ()

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
21 87
22 88
23 89
Fig. 2. Flowchart of the glide trajectory generation algorithm.
24 90
25 91
D 1 and D 2 are two parameters to be solved in the process
26 92
of parameterization, meeting the needs of maneuver and total
27 93
range. The other parameters are xed to simplify the calcula-
28 94
tion and speed up the solution. Considering this, articially set
29 95
( E 1 , E 2 , E 3 , E 4 ) to a constant value between 0 and 1 at regular in-
30 96
tervals to reduce the number of parameters to be solved.
31 97
The value of D f is determined by the desired terminal state,
32 98
and it is expressed as
33 99
34 D f = q f S r C D f /m (19) 100
35 101
36 where S r and m are the reference area and mass of the vehi- 102
37 cle, respectively. C D f is the drag coecient corresponding to the 103
Fig. 3. Drag acceleration prole with adjustable angle of attack where D 1 and D 2
38 are two adjustment parameters corresponding to the lateral maneuverability and maximum lift to drag ratio in terminal states. q f is the terminal 104
39 trajectory length. dynamic pressure, which can be computed as 105
40 106
1
41 constraints in (14), (15) depend only on the angle of attack . The qf = f v 2f (20) 107
42 maximum and minimum drag accelerations allowed in (14), (15) 2 108
43 are monotonically increasing with [10 , 20 ], which means The trajectory length corresponding to a line which combines 109
44 that we could get the maximum boundary of the longitudinal drag ( E a , D a ) and ( E b , D b ) in the drag prole can be expressed as: 110
45 entry corridor with = 20 and get the minimum boundary with 111
46 = 10 . The drag acceleration prole is represented as ve piece- E b 112
47 wise ane functions of the normalized non-conventional energy S ab = 1/ D ( E )dE (21) 113
48 E, as shown in Fig. 3. 114
49 Ea 115
uncertain = { D 1 , D 2 } (16)
50 116
Where
51 In order to simplify the generation of the drag prole, ( E 1 , E 2 , 117
52 E 3 , E 4 ) and D f are xed and ( E c , D c , E f ) could be calculated by Db Da 118
D(E ) = Da + (E Ea) (22)
53 the initial and terminal states. Eb Ea 119
54 The relationship between the actual energy E and the non- 120
55 dimensional energy E can be expressed as: Hence, the total trajectory length S can be described as the 121
function of two adjustment parameters D 1 and D 2 as:
E E0
56 122
57 E = (17) 123
58
E f E0
5
124
S (D1, D2) = Si (23)
59 where E 0 is the initial energy in the entry interface (EI) point, and 125
i =1
60 E f is the terminal energy. 126
61 In this paper, the values of ( E 1 , E 2 , E 3 , E 4 ) are set as follows: S is dened as the desired trajectory length, D 1 and D 2 are 127
62 solved as follows: 128
63 E 1 = 0.3 Step 1: Set initial values of D 1 and D 2 , and check the lateral 129
64 E 2 = 0.5 maneuverability of the vehicle in current condition. If it is large 130
(18)
65 E 3 = 0.7 enough to avoid the no-y zone, go to Step 2, otherwise, go to 131
66
E 4 = 0.8 Step 3. 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.5 (1-9)
R. He et al. / Aerospace Science and Technology () 5

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
21 87
22 88
23 89
24 Fig. 4. Lateral entry corridors corresponding to different drag acceleration proles. 90
25 91
26 Step 2: let D 1 = D 2 , both D 1 and D 2 are employed as the 3.2.2. Generation of the lateral prole 92
27 adjustment parameters to meet the requirement of S . A New- The maximum and minimum boundaries of lateral corridor are 93
28 ton iteration scheme is used to calculate the value of D 1 and D 2 , directly determined by the current drag acceleration prole. The 94
29 which is expressed as: QEGC condition can be expressed as: 95
 
 L
30 96
(k)
D1 D1
(k1)
 g v 2 /r  L
31
D2
(k+1)
= D1
(k+1)
= D1
(k) (k)
Sf S
(24) D= ,  sin  = 1 cos2 (28) 97
32 (k) (k1) ( L / D ) cos D D 98
Sf Sf
33 99
where r is the radius from the vehicle to the Earth center, which
34 (k) (k1) (k) 100
where D 1 = D 1 + D and D is a small positive constant. S f could be obtained by the approximation:
35
(k) (k)  
101
36 is the practical trajectory length corresponding to D 1 , D 2 , and
r r 2f 102
r02 r02 E 2 (29)
37 its obtained by the integration of the equations of motion. 103
38 Step 3: Adjust the value of D 1 by a xed increment to increase 104
E is the non-dimensional energy. As the drag acceleration is given
39 the lateral maneuverability, and D 2 is the only adjustment param- 105
in the generation of drag prole, for a given value of E [0, 1], we
40 eter, which is solved by the Newton iteration scheme to satisfy the 106
have
41 desired trajectory length S .  107
   2  2
42 Step 4: Check again whether the no-y zone is avoided success- L  L g v 2 /r 108
fully, if it is, stop the calculation, or return to Step 1.  sin  = (30)
43 D  D D
109
44 110
45 3.2. Lateral sub-planner Let be the angle of attack satisfying the desired drag acceler- 111
46 ation, min = max(10 , ). The lift-to-drag ratio L/D is monotonic 112
47 3.2.1. Lateral dynamics decreasing with [min , 20 ], we could get the upper and lower 113
48 The path angle is a small variable in the glide phase, let boundaries of ( L / D ) sin based on Eq. (30). 114
49 cos 1. r is assumed to be a constant value r (e.g., the aver- Fig. 4 shows two lateral entry corridors corresponding to dif- 115
50 age radial distance of the phase.) and a reduced order system is ferent drag acceleration proles. Furthermore, the lateral prole is 116
51 derived from Eqs. (4), (5), (7): designed as the interpolation result of the upper and lower bound- 117
52
    aries of lateral corridor, which can be expressed as: 118
53 d tan sin 1 1 L 119
= sin + C 1 + C 2 (25) L
54 dE r D V2 D sin = Lzmin + ( Lzmax Lzmin ) (31) 120
55   D 121
d sin 1
56 = (26) where Lzmax is the maximum boundary of the lateral entry corri- 122
57 dE r cos D dor, Lzmin is the minimal one, and is the weighted value between 123
 
58 d cos 1 0 and 1. In the lateral sub-planner, the weighted value is iterated 124
59 = (27) to minimize the cross-range error. 125
dE r D
60 In fact, the lateral corridor should include corridors within the 126
61 where C 1 , C 2 are two variables relating to the angular rate of positive and negative planes, corresponding to the positive and 127
62 planet rotation, shown in Eq. (7). As the drag acceleration prole negative values of bank angle, as shown in Fig. 5. 128
63 is given, at each value of energy E, the values of (, , ) de- The negative part of the lateral entry corridor below the x axis 129
64 pend only on the lateral prole variable ( L / D ) sin . It means that, is symmetrical to the upper one, which could be analyzed from 130
65 through the adjustment of ( L / D ) sin , the lateral motion of entry Eq. (28). Since the introduction of quasi-equilibrium glide condi- 131
66 could be generated effectively. tion, the maximum and minimum boundaries of lateral corridor 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.6 (1-9)
6 R. He et al. / Aerospace Science and Technology ()

1 Table 1 67
2 Calculation results of the prole parameters corresponding to three different entry 68
3 trajectories. 69
4 D1 D2 70
5 Trajectory 1 2.512 3.271 0.791 71
6 Trajectory 2 2.867 3.835 0.532 72
7 Trajectory 3 2.511 3.012 0.214 73
8 74
9 75
Due to the introduction of a series of assumptions in the prole
10 76
design process, such as the quasi-equilibrium glide condition, the
11 77
task of the tracker is to make the actual ight prole consistent
12 78
with the standard one. During the ight, the angle of attack and
13 79
Fig. 5. Boundaries of the lateral corridor. bank are constantly adjusted to ensure the requirements of down-
14 80
range and cross-range. Combing Eq. (32) and Eq. (35), we have
15
are directly determined by the longitudinal drag acceleration pro-   81
16


2 82
17
le. Only if the lateral prole is selected within the boundary, the L cos / D = a b + D r + 2 D + D + k1 DdE (37) 83
aircraft can be adjusted by the angle of attack and bank to achieve
18 84
the quasi-equilibrium gliding in current state. The upper and lower Combing Eq. (7) and Eq. (36), we have
19 85
20
boundaries of the lateral corridor describe the allowable range of  86
lateral control. The closer the lateral prole is to the upper bound- tan cos sin
21 L sin / D = v 2 cos C 1 C 2 87
ary, the stronger the lateral maneuverability of the aircraft. rD
22  88
23
89
3.3. Completion of the 3-DOF entry trajectory + re f + k2 + k3 dE (38)
24 90
25 91
In our previous study [32], it has been proven that there is only re f in Eq. (34) is used as the actual angle of attack. Let
26 92
one pair of control variables (re f , re f ) corresponding to the point A = L sin / D, B = L cos / D, C = ( L / D )re f and is the weighted
27 93
[D , ( L / D ) sin ] at the current energy. value between 0 and 1. Thus, the bank angle can be solved from
Since the ight path angle is so small that setting sin
28 94
= , Eq. (37) and Eq. (38), which takes the form:
cos
29 95
= 1, we have [32]  
30
 = sin1 ( A /C ) + (1 ) sgn( A )cos1 ( B /C ) (39) 96
31 ( L / D ) cos = a D

b (32) 97
32 4. Testing and discussion 98
33 where 99
34 a = h s v 2 In this study, the high performance Common Aero Vehicle 100
35 
2   (CAV-H) model is used in the simulations. The CAV-H has a refer- 101
C

D CD
2 C
D 4D
36
b=D +D + ence area of 0.48 m2 , and a mass of about 907 kg. The lift-to-drag 102
37 CD C 2D v2 CD v4 (33) ratio is monotonic increasing with [10 , 20 ] and has a maxi- 103
38  2 mum value of 3.5. The control constraints are set as [10 , 20 ], 104
v 1 C1 + C2
39 + g + [89 , 89 ], 5 /s and 20 /s. Other path constraints 105
40 r hs v 2 D hs are specied Q max = 1500 kw/m2 , qmax = 100 kPa, nmax = 3 g. 106
41 Combing Eq. (32) and the generated [D , ( L / D ) sin ], let L z = The mission scenario and path and terminal constraints are 107
42 ( L / D ) sin , we could get the control variables (re f , re f ): set as follows: entry trajectory starts from a preselected position 108
43
 (0 ,0 ), avoids a predetermined no-y zone, and then reaches a 109
44 L 2 nal position with desired energy. The radius of the no-y zone 110
(re f , Ma) = b
L 2z + a2 D

45 D is 1000 km and the center location is (35 ,1 ). All the path con- 111
46
  (34) 112
L straints are satised strictly in the whole entry. The initial altitude
47 sin re f = L z / (re f , Ma) h0 = 80 km, velocity v 0 = 6500 m/s, heading angle 0 = 60 and 113
D
48 path angle 0 = 0 . The desired TAEM interface point is chosen as 114
49 Furthermore, a tracker which follows both the longitudinal drag being 100 km away from the target with a nal altitude of 32 km 115
50 prole and lateral heading angle prole is proposed, based on the and velocity of 1965 m/s. 116
51 error dynamic model. In the longitudinal plane, the drag accelera- 117
52 tion error D = D re f D cur is taken as the state of interest. Con- 4.1. Entry trajectory generation with different terminal targets 118
53 sider the differential term, the proportional term and the derivative 119
54 term to build a second-order linear drag acceleration error dy- In this case, the adaptability of the trajectory planning algo- 120
55 namic, which is expressed as: rithm to different terminal targets is examined. A cylindrical no-y 121
56 zone is set in the path of the entry ight and three different termi- 122
57 D

+ 2 D
+ 2 D + k1 DdE = 0 (35) nal targets are (58 , 7.9 ), (51.6 , 7 ) and (62 , 6.9 ). Three 123
58 corresponding entry trajectories are obtained using the proposed 124
59 Similarly, in the lateral plane, the heading angle error = algorithm. The longitudinal drag acceleration prole and lateral 125
60 re f cur is taken as the state of interest, and a rst-order linear lift-to-drag prole are generated according to the lateral maneu- 126
61 heading angle error dynamic is achieved, which is expressed as: verability and trajectory length. Then, a tracker which follows both 127
62 the drag prole and heading angle prole is used to complete the 128
63
+ k 2 + k 3 dE = 0 (36) 3-DOF entry trajectory. The calculation results of the parameters in 129
64 the longitudinal and lateral proles are shown in Table 1. 130
65 where k1 , k2 and k3 are the coecients of the proportion part and Fig. 6 shows the reference and track results of the drag and 131
66 integration part [11], k1 = 1.05 105 , k2 = 0.18, k3 = 106 . heading angle proles corresponding to the three trajectories. In 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.7 (1-9)
R. He et al. / Aerospace Science and Technology () 7

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 Fig. 6. Track results of the drag and heading angle proles. 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
21 87
22 88
23 89
24 90
25 91
26 92
27 93
28 94
29 95
30 96
Fig. 7. Bank angle proles. Fig. 8. Flight path angle proles.
31 97
32 98
33 the initial phase, the altitude is relatively high and the aerody- Table 2 99
terminal states of entry trajectories.
34 namic force is too small. After a short adjustment, the aerodynamic 100
35 lift of the vehicle is large enough to satisfy the QEGC, and the glide No. v f (m/s) h f (km) f (deg) 101
36 phase begins. In the glide phase, the drag acceleration prole is 1 1960 33.3 0.6 102
37 represented as ve piecewise ane functions of energy. Two con- 2 1965 32.1 0.5 103
3 1973 31.6 0.8
38 stant parameters D 1 and D 2 of the drag prole are adjusted to 104
39 meet the requirement of trajectory length and lateral maneuver- 105
40 ability. As shown in Fig. 6, the tracker is employed to generate are not used. The highly constrained entry trajectory generation 106
41 the closed-loop prole by further modifying the control variables. mission is accomplished using an adjustable angle of attack and 107
42 In the longitudinal drag prole, the maximum tracking error is bank. Furthermore, as no reversal of bank angle exists in the whole 108
43 0.02 m/s2 , and in the lateral heading angle prole, the value is 1 . ight, it is easier to follow the control commands. Table 2 shows 109
44 The drag acceleration prole is designed as a piecewise linear the terminal states of the three entry trajectories. 110
45 function, which is continuous but not continuously differentiable. 111
46 This means that while the reference altitude and velocity are con- 4.2. Comparison with the traditional method 112
47 tinuous, the ight path angle and the bank angle will have jumps, 113
48 as shown in Eq. (37). In this paper, in order to avoid a large jump Compared with the traditional method in [32], two important 114
49 in the bank angle, the command change at the segmentation point improvements are proposed in this method: Firstly, the bank angle 115
50 is limited, as shown in Fig. 7. Furthermore, in the aforementioned is adjusted smoothly and continuously as the reversal logic of bank 116
51 algorithm design, it is assumed that the ight path angle is small angle is not used. Secondly, the angle of attack could be adjusted 117
52 and the altitude satises the analytical form as in Eq. (29). Mainly at any moment in the whole ight. 118
53 tracking the longitudinal drag acceleration and lateral heading an- As shown in Fig. 10, entry trajectories with no-y zone con- 119
54 gle, and the actual change of the ight path angle is small, shown straint are generated using two different kinds of methods. To the 120
55 as Fig. 8. The jump of the command will affect the tracking ef- traditional method in [32], reversal logic of bank angle is pro- 121
56 fect to a certain extent, resulting in a small deviation from the posed to avoid the no-y zone based on a lateral azimuth error 122
57 actual results and reference results. Updates of the 3D prole and deadband. When the lateral azimuth error is out of the deadband, 123
58 smoothing at the segmentation point will be the focus in the fu- the sign of bank angle is reversed. On the other hand, one rever- 124
59 ture research. sal of bank angle is assumed after the avoidance of no-y zone, 125
60 Fig. 9 shows the comparison of the reference and actual vari- and the reversal point is iterated to minimize the nal lateral er- 126
61 ables corresponding to the three entry trajectories. All these entry ror. The traditional method is a typical iterative method of control 127
62 trajectories are generated successfully, avoiding the no-y zone and parameters, with which the entry trajectory could be generated 128
63 reaching the nal TAEM interface point. rapidly. However, because of the parametric control model used in 129
64 Since the altitude varies slowly and no jumps exist, the QEGC this method, the maneuverability is greatly restricted in the ight. 130
65 assumption is established in the whole ight. For this control Whats more, with the reversal logic used in this method, ve re- 131
66 mode, nominal angle of attack and reversal logic of bank angle versals of bank angle exist in the whole ight, which would bring 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.8 (1-9)
8 R. He et al. / Aerospace Science and Technology ()

1 67
2 68
3 69
4 70
5 71
6 72
7 73
8 74
9 75
10 76
11 77
12 78
13 79
14 80
15 81
16 82
17 83
18 84
19 85
20 86
21 87
22 88
23 89
24 90
Fig. 9. Generated entry trajectories under no-y zone constraint.
25 91
26 92
27 93
28 94
29 95
30 96
31 97
32 98
33 99
34 100
35 101
36 102
37 103
38 104
39 105
40 106
41 107
42 108
43 109
44 110
45 111
46 112
47 113
48 114
49 115
Fig. 10. Comparison with the traditional method.
50 116
51 117
great challenges to the command tracking system. Considering this, 5. Conclusion
52 118
effective improvements are made in this study. Both the angle
53 119
54
of attack and bank are adjusted as the primary control variables. In this paper, the relationship between the longitudinal drag 120
55
The no-y zone is avoided effectively and the nal TAEM interface prole and lateral lift-to-drag prole is described, based on the 121
56 is reached while satisfying all the path constraints. Longitudinal QEGC assumption. Entry trajectory with no-y zone constraint is 122
57 drag prole and lateral lift-to-drag prole are both adjusted, ac- generated rapidly with the adjustment of both the angle of at- 123
58 cording to the desired maneuverability and the nal position of tack and bank. Compared with the traditional method in [32], the 124
59 the target. A PID tracker is used to follow the generation results proposed method has two advantages. The maneuverability of the 125
60 of 3-DOF entry trajectories. Simulation results demonstrate that, vehicle is greatly improved with an adjustable angle of attack. Con- 126
61 with this method, all the control variables vary smoothly. The an- trol commands are easier to be followed since no reversal of bank 127
62 gle of attack could be adjusted to meet the current longitudinal angle during the ight are executed. The approach is tested us- 128
63 and lateral constraints, and no bank reversals were required dur- ing the CAV-H model, and entry trajectories in different terminal 129
64 ing the entire ight. The entry trajectory is generated successfully, conditions can be successfully generated and satisfy complex path 130
65 avoiding no-y zone and reaching the nal position with desired constraint. It is our purpose in this paper to develop an algorithm 131
66 energy. for large cross-range entry trajectory generation under no-y zone 132
JID:AESCTE AID:4252 /FLA [m5G; v1.224; Prn:17/10/2017; 10:46] P.9 (1-9)
R. He et al. / Aerospace Science and Technology () 9

1 constraint, in which the angle of attack and bank are both adjusted [16] Z.F. Luo, H.B. Zhang, G.J. Tang, Patched corridor: a novel lateral logic for skip 67
2 according to the maneuverability and trajectory length. More im- entry guidance, J. Guid. Control Dyn. 37 (2014) 16511658, http://dx.doi.org/ 68
10.2514/1.G000441.
3 provement of the tracker and testing of the algorithm are required 69
[17] T. Wang, H.B. Zhang, L. Zeng, G.J. Tang, A robust predictorcorrector entry
4 in the future research. guidance, Aerosp. Sci. Technol. 66 (2017) 103111, http://dx.doi.org/10.1016/
70
5 j.ast.2017.03.010. 71
6 Conict of interest statement [18] M.L. Xu, K.J. Chen, L.H. Liu, G.J. Tang, Quasi-equilibrium glide adaptive guid- 72
7 ance for hypersonic vehicles, Sci. China, Technol. Sci. 55 (3) (2012) 856866, 73
http://dx.doi.org/10.1007/s11431-011-4727-z.
8 None declared. 74
[19] M. Sagliano, E. Mooij, S. Theil, Onboard trajectory generation for entry ve-
9 75
hicles via adaptive multivariate pseudospectral interpolation, J. Guid. Control
10 Acknowledgements Dyn. 40 (2) (2017) 465475, http://dx.doi.org/10.2514/6.2016-2115. 76
11 [20] W.B. Yu, W.C. Chen, Z.G. Jiang, et al., Analytical entry guidance based on 77
12 This work was supported by the National Natural Science Foun- pseudo-aerodynamic proles, Aerosp. Sci. Technol. 66 (2017) 315331, http:// 78
dx.doi.org/10.1016/j.ast.2017.03.018.
13 dation of China (61703409). 79
[21] Z.J. Shen, P. Lu, Onboard generation of three-dimensional constrained entry
14 trajectories, J. Guid. Control Dyn. 26 (1) (2003) 111121, http://dx.doi.org/ 80
15 References 10.2514/2.5021. 81
16 [22] Z.X. Liang, Z. Ren, Q.D. Li, J. Chen, Decoupled three-dimensional entry trajectory 82
17 [1] J.C. Harpold, C.A. Graves, Shuttle entry guidance, J. Astronaut. Sci. 27 (3) (1979) planning based on maneuver coecient, in: Proc IMechE, Part G, J. Aerosp. Eng. 83
239268. 231 (7) (2017) 12811292, http://dx.doi.org/10.1177/0954410016650906.
18 84
[2] A.J. Roenneke, A. Markl, Reentry control of a drag vs. energy prole, J. Guid. [23] P. Lu, Entry Trajectory Optimization with Analytical Feedback Bank Angle Law,
19 Control Dyn. 17 (5) (1994) 916920, http://dx.doi.org/10.2514/3.21290. AIAA-2008-7268, 2008, pp. 118, https://doi.org/10.2514/6.2008-7268. 85
20 [3] A.J. Roenneke, Adaptive On-Board Guidance for Entry Vehicle, AIAA- [24] E. Mooij, Adaptive Heat-Flux Tracking for Re-Entry Guidance, AIAA-2014-4142, 86
21 2001-37343, 2001, pp. 111, https://doi.org/10.2514/6.2001-4048. 2014, pp. 119, https://doi.org/10.2514/6.2014-4142. 87
[4] J.B. Wang, X. Qu, Z. Ren, Hybrid reentry guidance based on the online trajec- [25] K. Bollino, M. Ross, D. Doman, Optimal Nonlinear Feedback Guidance for Reen-
22 88
tory planning, J. Astronaut. 33 (9) (2012) 12171224, http://dx.doi.org/10.3873/ try Vehicles, AIAA-2006-6074, 2006, https://doi.org/10.2514/6.2006-6074.
23 89
j.issn.1000-1328.2012.09.005. [26] K. Bollino, High-Fidelity Real-Time Trajectory Optimization for Reusable Launch
24 [5] W. Grimm, J.G. van der Meulen, A.J. Roenneke, Optimal update scheme for drag Vehicles, Ph.D. Dissertation, Naval PostGraduate School, 2006. 90
25 reference proles in an entry guidance, J. Guid. Control Dyn. 26 (5) (2003) [27] M. Sagliano, E. Mooij, S. Theil, Adaptive disturbance-based high-order sliding- 91
26 695701, http://dx.doi.org/10.2514/2.5123. mode control for hypersonic entry vehicles, J. Guid. Control Dyn. 40 (2017) 92
[6] K.D. Mease, J.P. Kremer, Shuttle guidance revisited using nonlinear geomet- 521536, http://dx.doi.org/10.2514/1.G000675.
27 93
ric methods, J. Guid. Control Dyn. 17 (6) (1994) 13501356, http://dx.doi.org/ [28] M. Sagliano, Development of a Novel Algorithm for High Performance Reentry
28 10.2514/3.21355. 94
Guidance, Ph.D. Dissertation, University of Bremen, 2016.
29 [7] S. Bharadwaj, A.V. Rao, K.D. Mease, Entry guidance trajectory tracking law via [29] P. Lu, J. Hanson, Entry trajectory design for the X-33 vehicle, in: 22nd Atmo- 95
30 feedback linearization, J. Guid. Control Dyn. 21 (5) (1998) 726732, http:// spheric Flight Mechanics Conference, 1997. 96
dx.doi.org/10.2514/2.4318. [30] USAF, Air Force Handbook109th Congress, Department of the Air Force,
31 97
[8] P. Lu, Regulation about time-varying trajectories: precision entry guidance Washington, D.C., 2005.
32 illustrated, J. Guid. Control Dyn. 22 (6) (1999) 784790, http://dx.doi.org/ 98
[31] D. Zhang, L. Liu, Y.J. Wang, On-line reentry guidance algorithm with both
33 10.2514/6.1999-4070. path and no-y zone constraints, Acta Astronaut. 117 (2015) 243253, http:// 99
34 [9] G.A. Dukeman, Prole-Following Entry Guidance Using Linear Quadratic dx.doi.org/10.1016/j.actaastro.2015.08.006. 100
35 Regulator Theory, AIAA-2002-4457, 2002, pp. 110, https://doi.org/10.2514/ [32] Y. Xie, L.H. Liu, G.J. Tang, W. Zheng, Highly constrained entry trajec- 101
6.2002-4457. tory generation, Acta Astronaut. 88 (2013) 4460, http://dx.doi.org/10.1016/
36 102
[10] K.D. Mease, D.T. Chen, P. Teufel, H. Schonenberger, Reduced-order entry tra- j.actaastro.2013.01.024.
37 jectory planning for acceleration guidance, J. Guid. Control Dyn. 25 (2) (2002) 103
[33] Z.X. Liang, Q.D. Li, Z. Ren, Waypoint constrained guidance for entry ve-
38 257266, http://dx.doi.org/10.2514/2.4906. hicles, Aerosp. Sci. Technol. 52 (2016) 5261, http://dx.doi.org/10.1016/ 104
39 [11] A. Saraf, J.A. Leavitt, D.T. Chen, K.D. Mease, Design and evaluation of an acceler- j.ast.2016.02.023. 105
ation guidance algorithm for entry, J. Spacecr. Rockets 41 (6) (2004) 986996, [34] Z.X. Liang, S.Y. Liu, Q.D. Li, Z. Ren, Lateral entry guidance with no-y zone
40 106
http://dx.doi.org/10.2514/1.11015. constraint, Aerosp. Sci. Technol. 60 (2017) 3947, http://dx.doi.org/10.1016/
41 [12] Y.L. Zhang, K.J. Chen, L.H. Liu, et al., Entry trajectory planning based on 107
j.ast.2016.10.025.
42 three-dimensional acceleration prole guidance, Aerosp. Sci. Technol. 48 (2016) [35] Y. Xie, L.H. Liu, J. Liu, W. Zheng, Rapid generation of entry trajectories with 108
43 131139, http://dx.doi.org/10.1016/j.ast.2015.11.009. waypoint and no-y zone constraints, Acta Astronaut. 77 (2012) 167181, 109
44 [13] A. Saraf, J. Leavitt, M. Ferch, K.D. Mease, Landing footprint computation for http://dx.doi.org/10.1016/j.actaastro.2012.04.006. 110
entry vehicles, in: AIAA Guidance, Navigation, and Control Conference and Ex- [36] T.R. Jorris, R.G. Cobb, Three-dimensional trajectory optimization satisfying way-
45 111
hibit, AIAA-2004-4774, 2004, pp. 114. point and no-y zone constraints, J. Guid. Control Dyn. 32 (2) (2009) 551572,
46 [14] E. Mooij, Adaptive Lateral Flight Control for a Winged Re-Entry Vehicle, AIAA- 112
http://dx.doi.org/10.2514/1.37030.
47 2003-5322, 2003, pp. 112, https://doi.org/10.2514/6.2003-5322. [37] K.N. Zhang, W.C. Chen, Reentry Vehicle Constrained Trajectory Optimization, 113
48 [15] Z.J. Shen, P. Lu, Dynamic lateral entry guidance logic, J. Guid. Control Dyn. 27 AIAA-2011-2231, 2011, pp. 116, https://doi.org/10.2514/6.2011-2231. 114
(2004) 949959, http://dx.doi.org/10.2514/1.8008.
49 115
50 116
51 117
52 118
53 119
54 120
55 121
56 122
57 123
58 124
59 125
60 126
61 127
62 128
63 129
64 130
65 131
66 132

Вам также может понравиться