Chapter 1

Rocket

1.1 Overview
A missile is an intelligent, autonomous rocket that is intended to destroy its target or
object by carrying its payload to a predetermined location. The missile’s target, trajectory,
warhead, range, velocity, and launch platform are all taken into consideration during design.
System components of Missile

1. Targeting

2. Guidance system

3. Flight system

4. Engine

5. Warhead

Figure 1.1: Component of missile

1.1.1 Classification of Missiles

1.1.1.1 On the basis of Type:
1. Cruise Missile

2. Ballistic Missile

1
1.1.1.2 On the basis of Launch Mode
1. Surface-to-Surface Missile
2. Surface-to-Air Missile
3. Surface-to-Sea Missile
4. Air-to-Air Missile
5. Air-to-Surface Missile
6. Sea-to-Sea Missile
7. Sea-to-Surface Missile
8. Anti-Tank Missile

1.1.1.3 On the basis of Range

1. Short-range missiles, traveling less than 1,000 kilometers
2. Medium-range missiles, traveling between 1,000–3,000 kilometers
3. Intermediate-range missiles, traveling between 3,000–5,500 kilometers
4. Intercontinental ballistic missiles (ICBMs), traveling more than 5,500 kilometers

1.1.1.4 On the basis of Propulsion

1. Solid Propulsion
2. Liquid Propulsion
3. Hybrid Propulsion
4. Ramjet
5. Scramjet
6. Cryogenic

1.1.1.5 On the basis they are oriented toward the target,

1. Unguided Missiles
2. Guided Missiles

1.2 Phases of Flight

• Boost phase: is characterized by its rapid acceleration to terminal speed
• Cruise phase:is characterized by low to medium control action. The missile ap-
proaches the target, but is restrictive with large lateral accelerations to minimize in-
duced drag and thus minimize the flight time
• Terminal phase:is generally the last second(s) of flight. During this phase the per-
missible control gain is often chosen to be large in order to minimize miss distance

2
Chapter 2

Dynamics of Rocket

Modeling the process is a nontrivial component of any control challenge. Finding the
most straightforward mathematical explanation that can accurately forecast how the physical
system will react to each input is the goal. A rigid dynamic body’s motion can always be
expressed in terms of translational, rotational, and angular inclinations. The Newtonian-
Euler modeling approach is used to drive the mathematical model of a rocket. As we know
a rocket has six DOF of a rigid body - three for translation and three for rotation.

2.1 Coordinate systems

2.1.1 The Inertia Frame F i
An inertial frame is defined as one in which Newton’s law of inertia holds. The earth
coordinate is assumed to system be an inertial frame and has a flat x − y plane. This
coordinate frame is referred to as a NED reference frame.

Figure 2.1: Inertia frame

2.1.2 The Body Frame F b

Its origin is attached to the center of the mass of the missile. The roll axis ib is aligned to
the nose of the airframe, pitch axis j b points out the right wing, and yaw axis k b perpendicular
to the other two axes forming a right-angled coordinate system and pointing out toward the
earth.

3
 Figure 2.2: Body frame

The equivalent rotation matrix from the inertia frame to the body frame is given by:
 
Cθ Cψ Cθ Sψ −Sθ
b
Rib (ϕ, θ, ψ) = Rv2 v2
(ϕ)Rv1 (θ)Rvv1 (ψ) = Sϕ Sθ Cψ − Cϕ Sψ Sϕ Sθ Sψ + Cϕ Cψ Sϕ Cθ  (2.1)
Cϕ Sθ Cψ + Sϕ Sψ Cϕ Sθ Sψ − Sϕ Cψ Cϕ Cθ

when C represents the cosine function and S represents the sine function. And ϕ, θ, and ψ
are the Euler angles, (yaw, pitch, and roll)

Name Description
x Position of the aircraft along ii in F i
y Position of the aircraft along j i in F i
z Position of the aircraft along k i in F i
u Linear velocity along ib in F b
v Linear velocity along j b in F b
w Linear velocity measured along k b in F b
ϕ Roll angle with respect to inertia
θ Pitch angle with respect to inertia
ψ Yaw angle with respect to inertia
p Roll rate along ib in F b
q Pitch rate along j b in F b
r Yaw rate along k b in F b

Table 2.1: State variables for missile equations of motion

Model Assumptions

• The missile has a symmetrical structure

• The center of gravity (CG) of the missile coincides with the origin of the body coordi-
nate system

• The mass of fuel relative to the mass of the rocket is very small

4
2.2 Translational subsystem
Apply Newton’s second law on a body passing through translational motion can be given
as [1]:
dVe i X i
m( ) = Fk (2.2)
dt k

Inertial Position(x,y,z)
In this section, the equation of the inertial position of a fixed-wing UAV is expressed. New-
ton’s second law of motion is used as a guiding rule
X X
f orcenet = f orceapplied − f orceretarding (2.3)

mV˙e = Rbi fpb + Rbi fab + fgi (2.4)

where Ve is the linear velocity of the aircraft in the inertia frame, fp is a propulsive force on
the body frame, fa aerodynamic force on the body frame, fg gravitational force on inertia
frame and Rbi is a rotational matrix of body frame to inertia frame.
        
ẍ Cθ Cψ Sϕ Sθ Cψ − Cϕ Sψ Cϕ Sθ Cψ + Sϕ Sψ fp 0 d
         x
m ÿ  =  Cθ Sψ Sϕ Sθ Sψ + Cϕ Cψ Cϕ Sθ Sψ − Sϕ Cψ   0  +  0  + dy  (2.5)
        
        
z̈ −Sθ Sϕ Cθ Cϕ Cθ 0 mg dz
   
dx fb
   ax 
dy  = Rbi fab = Rbi fay (2.6)
   b

   
b
dz faz
After rearranging equation 2.5 yields

f p dx
ẍ = cθcψ + (2.7)
m m
f p dy
ÿ = cθsψ + (2.8)
m m
f p dz
z̈ = −sθ + +g (2.9)
m m
Angular position     
ϕ̇ 1 tanθsinϕ cosϕtanθ p
    
 θ̇  = 0 −sinϕ  q  (2.10)
    
cosϕ
    
ψ̇ 0 sinϕ/cosθ cosψ/cosθ r

ϕ̇ = p + qsin(ϕ)tan(θ) + rcos(ϕ)tan(θ) (2.11)

θ̇ = qcos(ϕ) − rsin(ϕ) (2.12)

ψ̇ = qsin(ϕ)sec(θ) + rcos(ϕ)sec(θ) (2.13)

Body frame translation velocity
dVgb X
m( b
+ wb/i × Vgb ) = Fb (2.14)
dtb k

5
     P   
u p (fxa + fxp + fxg = Fx ) u̇
    P   b
 
 dV
When Vgb =  v  ,wb/i =  q  , k F b =  (fya + fyp + fyg = Fy )  , dtbg =  v̇ 
  b   P  
    P   
w r, (fza + fzp + fzg = Fz ) ẇ
By computing the cross product in equation 2.14 we can obtain the following expressions.
     
u̇ rv − qw F
     x
 v̇  = pw − ru + 1/m Fy  (2.15)
     
     
ẇ qu − pu Fz

1
fxa = − ρVM 2 SCA (2.16)
2
1
fya = ρVM 2 SCN y (2.17)
2
1
fza = − ρVM 2 SCN z (2.18)
2
where CA represents aerodynamic axial force coefficient, dimensionless, CN y represents co-
efficient corresponding to component of normal force on body frame y-axis, CN z represents
coefficient corresponding to component of normal force on body frame z-axis, S represents
aerodynamic reference area, ρ represents atmospheric density and VM represents magnitude
of velocity vector of the center of mass of the rocket.

2.3 Rotational subsystem

In the case of rotational motion Newton’s Law is stated as [1]:

dh i X i
( ) = mk (2.19)
dt k

A gyroscope or gyro is a device that measures the angular acceleration or rotational motion
of a dynamic body. On a rocket, this rotational motion can be described as
X
b
ẇb/i = J −1 (−wb/i
b b
× (Jwb/i )+ mb ) (2.20)
k
 T
By letting b
wb/i =
p q r be the angular rates and the summation of the moment be
 
km = L M N
b T
P

Where L, M , and N roll moment, pitch moment, and yaw moment respectively.
Finally, the rotational dynamics can be written as:
   
La +Lp −qr(Izz −Iyy )
ṗ Ixx
   
(2.21)
b   Ma +Mp −rp(Ixx −Izz ) 
ẇb/i = q̇  = 

Iyy

   
Na +Np −qp(Iyy −Ixx )
ṙ Izz

where La , Ma and Na are components of aerodynamic moment in body frame, Lp , Mp and

Np are components of propulsion moment in body frame and Ixx ,Iyy , Izz are components of

6
inertia (diagonal elements of inertia matrix when products of inertia are zero) in this work
for simplicity assume inertia compponents are equal.

1
La = ρVM 2 Cl Sd (2.22)
2
1
Ma = ρVM 2 Cm Sd (2.23)
2
1
Na = ρVM 2 Cn Sd (2.24)
2
when Cl is aerodynamic roll moment coefficient about center of mass, Cm is aerodynamic
pitch moment coefficient about center of mass, Cn is aerodynamic yaw moment coefficient
about center of mass and d is aerodynamic reference length of body. The aerodynamic
moment coefficients are obtained as
d
C l = C l δ δa + (Clp p) (2.25)
2VM

xcm − xref d
Cm = Cmref − CNz + (Cmq + Cmα )q (2.26)
d 2VM
xcm − xref d
Cn = Cnref + CNy + (Cnr + Cnβ )r (2.27)
d 2VM
And
Cmref = Cmα α + Cmδ δe (2.28)

Cnref = Cnβ β + Cnδ δr (2.29)

where Clp is roll damping derivative relative to roll rate, Clδ is slope of curve formed by roll
moment coefficient, is pitching moment coefficientCmref , Cmq is pitch damping derivatives,Cmα
is pitch damping derivative relative to angle of attack,xcm is instantaneous distance from
rocket nose to center of mass, xref is distance from rocket nose to reference moment station,
α is angle of attack, β is side sleep angle and δa , δe , δr are aileron , elevator and rudder
deflection surface respectively.

7
2.4 Control surface(Fin),angle of attack and side sleep
angle

Figure 2.3: Control surface

We can express δa , δe and δr as a function of δ1 , δ2 , δ3 and δ4 .

    
δa −1 −1 −1 −1 δ1
    
=
δe  −1 −1 1 (2.30)
    
1  δ2 
    
δr 1 −1 −1 1 δ3
√
VM = u2 + v 2 + w 2 (2.31)

α = tan−1 (w/v) (2.32)

√
β = sin−1 (v/ u2 + v 2 + w2 ) (2.33)

2.5 Decoupling
The primary goal of this work to control the inertial position (x, y, z), and attitude
(ϕ, θ, ψ). The mathematical models of rockets are still very complex and highly coupled.
To simplify the control process, it is necessary to decouple the mathematical models before
designing the controllers. Decoupling processing mainly extracts the dominant state quantity
in the control process and treats the remaining control quantity as uncertainty

• Assumption In most flight time ϕ,θ,ψ will be very small.

Roll Angle
ϕ̇ = p + qsin(ϕ)tan(θ) + rcos(ϕ)tan(θ) (2.34)

ϕ̇ = p + dϕ1 (2.35)
and considering dϕ1 as a disturbance and differentiate then substitute ṗ

dϕ1 = qsin(ϕ)tan(θ) + rcos(ϕ)tan(θ) (2.36)

8
 rocket.PNG

Figure 2.4: Control inputs and Outputs of Rocket

ϕ̈ =
1 ˙ ) + 1 ρV 2 SdCl δa + Iyy − Izz qr
ρd2 SVM Clp (ϕ̇ − dϕ1 (2.37)
2Ixx 2Ixx M δ
Ixx
ϕ̈ = aϕ1 ϕ + aϕ2 δa + dϕ2 (2.38)

ϕ̈ = aϕ1 ϕ + aϕ2 δa + dϕ2 (2.39)

Pitch angle
θ̈ = q̇ + d˙θ1 (2.40)
1 1 1
θ̈ = ρSdCmα θ̇ + ρVM Sd2 (Cmq + Cmα )θ + ρVM2 SdCmδ δe + dθ2 (2.41)
2Iyy 4Iyy 2Iyy

θ̈ = aθ1 θ̇ + aθ2 θ + aθ3 δe + dθ2 (2.42)

Yaw Angle
ψ̇ = qsin(ϕ)sec(θ) + rcos(ϕ)sec(θ) (2.43)

ψ̇ = r + dψ1 (2.44)
where dψ1 the disturbance
1 1
ψ̈ = ρVM Sd2 (Cnr + Cnβ )ψ + ρV 2 SdCnδ δr + dψ2 (2.45)
4Izz 2Izz M

9
 ψ̈ = aψ1 ψ̇ + aψ2 δr + dψ2 (2.46)

The
 six-state decoupled second order equation of motion of Rocket

 ϕ̈ = aϕ1 ϕ + aϕ2 δa + dϕ2

θ̈ = aθ1 θ̇ + aθ2 θ + aθ3 δe + dθ2





ψ̈ = a ψ̇ + a δr + d
ψ1 ψ2 ψ2
fp d


 ẍ = cθcψ m + m x

ÿ = cθsψ fmp + dmy






z̈ = −sθ fmp + dmz + g


In the above equation dϕ2 , dθ2 , dψ2 , dmx , dmy and dmz + g are considered as the uncertainty.

2.5.1 State space representation

f p dx dx fp
ẍ = cos(θ)cos(ψ) + = Ux + where Ux = cos(θ)cos(ψ) (2.47)
m m m m

x1 = x, x2 = ẋ

x˙1 = x2 (2.48)

x˙2 = Ux + dmx


f p dy dy fp
ÿ = cos(θ)sin(ψ) + = Uy + where Uy = cos(θ)sin(ψ) (2.49)
m m m m

x3 = y, x4 = ẏ

x˙3 = x4 (2.50)
x˙4 = Uy + dmy



f p dz dz fp
z̈ = −sin(θ) + + g = Uz + +g where Uz = −sin(θ) +g (2.51)
m m m m

x5
 = z, x6 = ż
x˙5 = x6 (2.52)

x˙6 = Uz + dmz


ϕ̈ = aϕ1 ϕ̇ + aϕ2 δa + dϕ2 (2.53)


x7 = ϕ, x8 = ϕ̇

x˙7 = x8 (2.54)

x˙8 = aϕ1 x8 + aϕ2 δa + dϕ2


θ̈ = aθ1 θ̇ + aθ2 θ + aθ3 δe + dθ2 (2.55)


x9 = θ, x10 = θ̇

x˙9 = x10 (2.56)

x˙10 = aθ1 x10 + aθ2 x9 + aθ3 δe + dθ2


ψ̈ = aψ1 ψ̇ + aψ2 δr + dψ2 (2.57)


x11
 = ψ, x12 = ψ̇
x˙11 = x12 (2.58)

x˙12 = aϕ1 x12 + aϕ2 δr + dϕ2


10
2.6 Guidance
Missile Guidance The term "missile guidance" describes a range of techniques used to
direct a missile or guided bomb to its desired target. The accuracy of the missile’s target is
a crucial component of its potency. Guidance systems increase the probability of guidance,
which increases missile accuracy. [2].

Figure 2.5: Caption

The guidance law are classified in to two main groups.

1. Acceleration based guidance laws

2. Angle based guidance laws.

2.6.1 Acceleration-based guidance laws

An acceleration control system is designed in order to obey the guidance commands.
Proportional navigation guidance (PNG), velocity pursuit guidance (VPG) and augmented
proportional navigation guidance (APNG) laws are acceleration based guidance law [3].

2.6.1.1 Proportional navigation guidance law

As the most widely used guidance law for missiles, the PNG law can be expressed re-
garding the yaw and pitch planes of the missile–target.

2.6.1.2 Velocity pursuit guidance law.

In this approach, the goal is to align the missile velocity vector VM and line-of-sight
vector rM and thus to keep the direction of VM as indicating the target [4].

2.6.1.3 Augmented proportional navigation guidance law

In this scheme, the guidance commands to the lateral acceleration components of the
missile are determined by adding the product of the half of the relevant lateral acceleration
component of the target by the corresponding effective navigation ratio to the command
expressions derived for the PNG law.

2.6.2 Angle-based guidance laws

As the most commonly used angle-based guidance laws, the body pursuit guidance (BPG)
and linear homing guidance (LHG) laws

11
2.6.2.1 Body pursuit guidance law.
In the BPG approach, it is intended to put the longitudinal axis of the missile, that is,
the u axis, onto the line-of-sight between the missile and target.

2.6.2.2 Linear homing guidance law.

In the LHG approach, the most appropriate way is to orient the missile velocity vector
towards the predicted intercept point at which the missile–target collision will occur after a
while. Then, the resulting guidance commands will be in the form of the flight path angles
of the missile in the yaw and pitch planes

2.7 Robust Direct MRAC Design In the Absence of d(t)

Many of the rocket’s physical parameters depend on different operating conditions. For
outdoor operations, this effect is significantly influenced by temperature, pressure, and many
other unknown scenarios of the environment, due to this reason, uncertainty occurs during
the flight period. In these cases, conventional fixed gain feedback controllers fail to ensure
stabilization and conventional MRAC causes instability in the presence of unmatched uncer-
tainty due to disturbance. Hence, Robust adaptive control techniques are designed so that
controller gains can be tuned online to handle the variation and the disturbance

2.7.1 Control Architecture

As we know rocket is under actuated system it has four inputs and six outputs, there-
fore to controlling it directly is impossible. The overall control system is designed as the
outer(position) and inner(attitude) loop. First, the state feedback model reference adap-
tive controller is designed for position control. To be able to control x, y, and z positions
indirectly, an outer loop is used. Outer loop controller outputs Ux ,Uy and Uz are used to
determine the desired pitch angle (θd ) and yaw angle (ψd )(conversion block). Second, the
inner loop model reference adaptive controller is designed to track the command input angles
θr , ψr and ϕr . The appropriate ϕr is set for given trajectories.
Plant dynamics with matched parametric and unmatched uncertainty are given as

ẋ = Ax(t) + B(u(t) + Θ∗T ϕ(x)) + d(t) (2.59)

(A, B) is controllable x(t)∈ Rn is a state vector, A is system matrix A ∈ Rn xRn , B is
input matrix B∈ Rn xRm , U (t) ∈ Rm is a control input, Θ∗ ∈ Rp xRn is parametric uncer-
tain matrix, ϕ(x) is a known bounded regressor function. The quantity Θ∗ T ϕ(x) is called
a parametric matched uncertainty, and d(t) is unmatched uncertainty due to disturbance,
and it is bounded, but in this work, d(t) represent some part of dynamics and considered as
uncertainty [5]. When a parametric uncertainty is matched, the control input can cancel out
the uncertainty completely when the adaptation is perfect. In this thesis ϕ(x) =x, when x is
the state of the plant ϕ or θ or ψ. The parametric uncertain matrix and regressor function
are expressed as
The reference model which is designed based on the desired performance specification for
second order system is given by

x˙m = Am xm (t) + Bm r(t) (2.60)

Where xm (t) is a uniformly bounded model reference signal, r(t) is a piecewise continuous
bounded reference command signal and Am is Hurwitz

12
 rocket_arch.PNG

Figure 2.6: Control Architecture

Then the ideal controller in nominal case

u∗ = kx∗T x + kr∗ r − Θ∗T ϕ(x) (2.61)

Where u∗ is ideal control input kx∗ , kr∗ and Θ∗T are final constant estimated values of con-
troller gains when the plant model output is perfectly track the reference model response.
Model matching condition

A + Bkx∗T = Am
Bkr∗ = Bm
The actual full-state feedback adaptive controller is designed as:

uad = kx (t)T x(t) + kr (t)r(t) − Θ(t)T ϕ(x) (2.62)

Where kx (t)∈ R2 ,kr (t)∈ R1 , and Θ(t)∈ R2 are parameters to be estimated on-line in this
case uad represent ux or uy or uz or δa or δe or δr
Estimation error are given by

kˆx = kx (t) − kx∗ , kˆr = kr (t) − kr∗ , Θ̂ = Θ(t) − Θ∗ (2.63)

Substitute equation 2.62 into the plant model 2.59 and d(t) is neglected. Then the closed-loop
dynamics are given by
T
ẋ = (A + Bkx∗T + B kˆx )x + (Bkr∗ + B kˆr )r − B Θ̂T ϕ(x) (2.64)

13
Now compute the closed-loop tracking error and differentiate the error dynamics

e(t) ˙ = Am e − B kˆx T x − B kˆr r + B Θ̂T ϕ(x)

˙ = xm˙(t) − x(t) (2.65)
choose the Lyapunov candidate function, it is a positive definite function
T
V (e, kˆx , kˆr , Θ̂) = eT pe + trace(kˆx Γ−1 ˆ T −1 T −1
(2.66)
x K̂x + kr Γr K̂r + Θ̂ ΓΘ Θ̂)

Where Γx =ΓTx >0∈ R2 xR2 ,Γr =ΓTr >0∈ R1 xR1 ,ΓΘ =ΓTΘ >0∈ R2 xR2 are called the adaptation
(learning) rates matrix for x, r and Θ respectively. For Hurwitz/asymptotically stable Am

P Am + ATm P = −Q (2.67)

Where P = P T > 0 and Q = QT > 0, the time derivative of V is evaluated along 2.65 trajec-
tory.
T ˙ ˆ T −1 ˙ T −1 ˙
V̇ = e˙T P e + eT ė + 2tr(kˆx Γ−1
x K̂x + kr Γr K̂r + Θ̂ ΓΘ Θ̂)
T
= (Am e − B kˆx x − B kˆr r + B Θ̂T ϕ(x))T P e
T
+ eT P (Am e − B kˆx x − B kˆr r + B Θ̂T ϕ(x))
T ˙ ˆ T −1 ˙ T −1 ˙
+ 2tr(kˆx Γ−1
x K̂x + kr Γr K̂r + Θ̂ ΓΘ Θ̂) (2.68)
Substitute equation 2.67 into 2.68 yields
 
T −1 ˙
T T ˆT ˆ
V̇ = −e Qe + −2e P B kx x + 2tr(kx Γx K̂x )
 
T −1 ˙
T ˆ
+ −2e P B kr r + 2tr(kˆr Γr K̂r )
T

 
˙
+ 2e P B ΘˆT ϕ(x) + 2tr(Θ̂ ΓΘ Θ̂)
T T −1
(2.69)

Using vector trace identity

eT P B kˆxT x = tr(kˆxT xeT P B)

eT P B kˆT r = tr(kˆT reT P B)
r r

e P B ΘˆT ϕ(x) = tr(ΘˆT ϕ(x)eT P B)

T
(2.70)
Substitute equation 2.70 into 2.69 yields
 
T ˆ −1 ˙ T
V̇ = −e Qe + 2tr(kxT Γx K̂x − xe P B )
 
˙
+ 2tr(kˆrT Γr K̂r − re P B )
−1 T

 
ˆ −1 ˙
T T
+ 2tr(Θ ΓΘ Θ̂ + ϕ(x)e P B ) (2.71)

V̇ = −eT QeV̇ ≤ 0(globally negative semi-definite). If the adaptive laws are selected as
˙
k˙x = kˆx = Γx x(t)eT P B
˙
k˙r = kˆr = Γr r(t)eT P B
˙
Θ̇ = Θ̂ = −ΓΘ ϕ(x)eT P B (2.72)

14
2.7.2 Robust MRAC Design In The Presence of Dynamics(d(t))
Which Is Considered as uncertainty
Consider the plant model given below

ẋ = Ax(t) + B(u(t) + Θ∗T ϕ(x)) + d(t) (2.73)

• Parameter Drift is a consequence when an adaptive parameter diverges when the
time elapsed in the presence of unmatched uncertainty due disturbance [6] and pa-
rameter drift is the result of the lack of a mathematical guarantee of convergence of
adaptive parameters. It leads to instability and sudden failure of the system.

Assume the system is operating in the presence of uncertainty due to disturbance d(t) ∈ (R)1

||d(t)|| ≤ d¯ (2.74)
¯
Constant upper bound d≥0, and the error dynamics are given as

e(t) ˙ = Am e − B kˆx T x − B kˆr r + B Θ̂T ϕ(x) − d(t)

˙ = xm˙(t) − x(t) (2.75)
A radially unbounded quadratic Lyapunov function candidate is selected in the familiar form
T
V (e, kˆx , kˆr , Θ̂) = eT pe + trace(kˆx Γ−1 ˆ T −1 T −1
(2.76)
x K̂x + kr Γr K̂r + Θ̂ ΓΘ Θ̂)

Suppose that we use the same adaptive update laws as the above subsections, that is,
˙
kˆx = Γx x(t)eT P B
˙
kˆr = Γr r(t)eT P B
˙
Θ̂ = −ΓΘ ϕ(x)eT P B (2.77)

Then
V̇ = −eT Qe − 2eT P d(t) (2.78)
Now determine the upper bound of 2.78

V̇ ≤ −λmin (Q)||e||2 + 2||e||λmax (P )d¯ (2.79)

• Case 1
2λmax (P ) ¯
||e|| ≥ d
λmin (Q)
In these case V̇ ≤0 (Negative semi-definite) ,and also all V (t), e(t), kˆx , kˆr and Θ̂ are bounded

• Case 2
2λmax (P ) ¯
||e|| ≤ d
λmin (Q)
In these case V̇ ≤ some positive number we can not say V (t), kˆx , kˆr and Θ̂ are bounded.

15
2.7.2.1 σ-Modification
The σ robust modification technique was proposed ed by Ioannu and Kokotovic [7]. A
number of modification techniques were proposed in the 1980s, among them the σ modifi-
cation is the simplest modification method that improve the robustness of uncertain plants.
The adaptive law with the σ-modification is given as

k˙x = Γx (x(t)eT P B − σx kx )
k˙r = Γr (r(t)eT P B − σr kr )
Θ̇ = −ΓΘ (ϕ(x)eT P B + σΘ Θ) (2.80)
where σx ,σr and σΘ are >0 is the modification parameter. These modification parameters
must chosen properly. The larger value of σx , σr , and σΘ increase the tracking errors, but
it offers better robustness. There is always a trade-off between robustness and asymptotic
tracking. Necessarily σ-modification term adds a constant damping term in the updated
law, which provides a technique to bound the adaptive parameters kx , kr and Θ.

2.7.3 Reference Model

The reference model should chosen by considering two basic requirements. The first one
is, that it should reflect performance specifications in the control tasks in terms of transient
response like settling time, rise time, and maximum overshoot. The second one is, that the
reference model should reflect the order, and relative degree of the plant model [8]. Since the
reference model sets the desired performance of the closed-loop system for a reference input,
it must be chosen properly by checking all the above criteria, which is the basic process in the
selection of a reference model for the MRAC system. By considering the above requirement,
a reference model is designed as follows
The reference model transfer function is given by
wn2
xm = 2 r (2.81)
s + 2ζwn s + wn2
wn the undamped natural frequency of a second-order system refers to the oscillation fre-
quency of the system in the absence of damping. ζ the damping ratio of a second-order
system quantifies the system’s resistance to changes in the output, indicating the degree of
dampening. wn and ζ can be determined from second-order specification settling time and
overshoot
• Settling time is the duration it takes for the response curve to reach and consistently
remain within a specified range around the final value, specified by absolute percentage
of the final value (usually 2% )
4
ts = (2.82)
ζwn
• Maximum Overshoot represents the highest peak value of the response curve, mea-
sured from unity.
√−ζπ
%M.O = e 1−ζ2 (2.83)
In a nested control approach, it is crucial to consider the system’s speed or response time.
This speed is responsible for determining the system’s ability to respond to changing input
commands or controlled parameters. To ensure overall closed-loop stability in the nested
control architecture, the response time of the inner loop must be faster than that of the
outer loop. In cases where servo or tracking systems are involved, the inner loop’s response
time should be faster than the outer loop’s.

16
2.8 Pre-defined Projectile Trajectory Generation
In general, a rocket can be guided by using a path-planning algorithm. It is important to
possess a path-planning algorithm that can effectively and efficiently calculate the optimal,
safest, and quickest route to the final destination. This en- sure that rocket arrives at’s
desired destination in a timely and secure manner. Which use waypoints to generate an
optimal trajectory. As we know projectile trajectory is an optimal trajectory that requires
minimum energy to hit the target.
Here, select short-distance missile
Range set to be=1000km
Maximum height set to be=100km
Based on the above-given data compute the initial speed, projection angle, and time taken
to complete the trajectory
The general formula for maximum height is given by:

vi2 sin(θ)2
zmax = hmax = (2.84)
2g
The general formula for maximum range is given by:

vi2 sin(2θ) 2v 2 sin(θ) cos(θ)

R= = i (2.85)
g g
The general formula for the time required to complete the trajectory is given by:

2vi sin(θ)
T = (2.86)
g
Where vi represents initial speed, θ represents angle of projection and g represents gravity
Now divide equation 2.84 by 2.85
hmax tan(θ)
= (2.87)
R 4
hmax
θ = arctan(4 ) = 21.8degree (2.88)
R
θ=21.8 degree, vi =3771.8 Am/s, T =285.5709 second The project motion is obtained at the
vertical plane

x(t) =vix t (2.89)

y(t) =0 (2.90)
1
z(t) =viy t − gt2 (2.91)
2
where vix and viy represent initial horizontal velocity and vertical velocity respectively and
t represent instantaneous time.

17
 Figure 2.7: Projectile trajectory

2.9 Result
To simulate the rocket system the aerodynamic coefficient is necessary, but here the
coefficients are not available.

18
Bibliography

[1] Randal W Beard and Timothy W McLain. "Small unmanned aircraft: Theory and
practice". Princeton University Press, 2012.

[2] James N Constant. Fundamentals of strategic weapons: offense and defense systems. 1,
1981.

[3] Bülent Özkan, M Kemal Özgören, and Gökmen Mahmutyazıcıoğlu. Modeling of dynam-
ics, guidance, and control systems of air-to-surface missiles. The Journal of Defense
Modeling and Simulation, 9(2):101–112, 2012.

[4] Bülent ÖZKAN, M Kemal Özgören, and Gökmen Mahmutyazicioğlu. Comparison of

the linear homing, parabolic homing and proportional navigation guidance methods on
a two-part homing missile against a surface target. Gazi University Journal of Science,
23(1):81–88, 2010.

[5] Nhan T. Nguyen. "model-reference adaptive control". Springer, pages 95–96, AG 2018.

[6] Nhan T. Nguyen. "model-reference adaptive control". Springer, AG 2018.

[7] Annaswamy A.M. Narendra, K.S. "a new adaptive law for robust adaptive control with-
out persistency of excitation". IEEE Trans.Automat Contr., Feb. 1987.

[8] Muluken Menebo. "neural network based model reference adaptive control of quadrotor
uav for precision agriculture". MSc Thesis, 2022.

19