Nghiên cứu xây dựng thuật toán nâng cao chất lượng xử lý tín hiệu kênh cao cho một lớp tên lửa hành trình đối hải trên cơ sở lý thuyết điều khiển hiện đại tt tiếng anh

MINISTRY OF EDUCATION AND TRAINING

MINISTRY OF NATIONAL DEFENSE

ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY

PHAM DUC THOA

RESEARCH ON THE CONSTRUCTION OF AN ALGORITHM FOR
IMPROVING THE QUALITY OF THE PROCESS OF HEIGHT MEASUREMENT
SIGNALS FOR A CLASS OF MARINE CRUISE MISSILES BASED
ON THE MODERN CONTROL THEORY

Major: Control Engineering and Automation
Code : 9 52 02 16

SUMMARY OF TECHNICAL DOCTORAL DISSERTATION

HA NOI – 2019


LIST OF PUBLISHED SCIENTIFIC WORKS

The thesis was completed at
ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY
1.

Supervisors:
1. Dr. Nguyen Quang Vinh

2.

2. Dr. Nguyen Xuan Can

Review 1: Assoc. Prof. Dr Pham Trung Dung

3.

Military Technical Academy
Review 2: Prof. Dr Nguyen Doan Phuoc
Hanoi University of Science and Technology

4.

Review 3: Assoc. Prof. Dr Tran Duc Thuan
Academy of Military Science and Technology
5.

The dissertation was defended in front of the Doctoral Evaluating Committee
at Academy level held at Academy of Military Science and Technology
at …/…, 2019

More information on the dissertation can found from:
- Academy of Military Science and Technology
- National Library

Pham Duc Thoa, Nguyen Quang Vinh, Nguyen Xuan Can, (2016),
”Building an algorithm for information processing in the
compound altimeter of a flying vehicle”, Journal of Military Science and
Technology, Academy of military science and technology, (9/2016) p.116122.
Pham Duc Thoa, Nguyen Quang Vinh, Nguyen Xuan Can, Tran Ngoc

Huong, (2018), “Application of the self-organnized algorithm
for improving the signal processing quality of the linked high measuaring
system”, Confference: “Apply high technology into practice”, Journal of
Military Science and Technology, Academy of military science and
technology, (8/2018) p.382-390.
Nguyen Quang Vinh, Pham Duc Thoa, (2018), “Improving high quality in
combination processing the high measurement signals”, The 7th International
Conference on Frontiers of Intelligent Computing: Theory and Applications
(FICTA2018); 29-30 November.
Pham Duc Thoa, To Ba Thanh, Nguyen Quang Vinh, Bui Minh Tuan,
(2019), “The construction of a self-organizing algorithm for choosing a
model of extrapolation in the combined process of signal of the height
measurement”, Journal of Military Science and Technology, Academy of
military science and technology, (3/2019), p269-278.
Pham Duc Thoa, Nguyen Quang Vinh, Tran Ngoc Huong, (2019),
“Evaluating the influence of the observability level to the exactness of
processing signals in the combination of the inertia height meter and the radio
height meter”, Journal of Military Science and Technology, Academy of
military science and technology, (4/2019), p.73-80.


1

PREFACE
1. The necessity of the dissertation
In terms of current combat operations, Electronic combat systems
have superior features such as high combat capability both in mobility,
the ability to suppress the operation of the system to lead the enemy in
the form of different types of noise. On the lead of many modern cruise
missiles equipped with inertial navigation system (INS), due to large

cumulative errors, from there, this error will cause a large error in the
control.
Due to the characteristics of cruise missiles with the condition of low
orbit in many altitude ranges, long flight times, under the influence of
different types of noise, changing some special parameters corresponding
to each combination of high measurement. Resulting in high quality
signal processing when combining each high measuring set with inertial
height measuring kit at each flight condition is different. The dissertation
“Research on the construction of an algorithm for improving the quality
of the process of height measurement signals for a class of marine cruise
missiles based on the modern control theory” studying the solution of
combining high-measuring set and signal processing algorithm to
optimize the structure of the high-measurement combination, ensuring
that the height information of cruise missiles is continuous and accurate;
Construction of an extrapolation model in the combined process of height
measurement by using a self-organization algorithm.
2. Research targets for the dissertation
Applying modern control theory on the basis of self-organizing
algorithm (SOA) and standard of observing the observed level of
variables in the state space to build intelligent high-quality measurement,
improve quality in the combination of high measurement signals.
3. Some main contents of the dissertation
- Construction of an algorithm for the combined process of height
measurement signals and choice of the structure of the combined height
meter based on the evaluation of the observably level
- Construction of an extrapolation model in the combined process of
height measurement by using a self-organization algorithm with the


2

condition of an in consonant observably level of status variable
4. Object and scope of the dissertation research
Modern high-altitude measuring system in the high-channel control
system of a class of maritime cruise missiles.
5. Approaches of study
Combining method of theoretical research and using simulation
techniques to test and evaluate algorithms.
6. Scientific and practical benefits of this project
Use the standard of observable level and SOA to build an
extrapolation model to optimize signal processing in high measurement
combination. Overcoming the limitations that the Kalman filter cannot
solve during processing combined with high measurement signals.
The results of the thesis can be used for designing and improving the
control system-stabilizing the height for a class of cruise missiles, adding
methodologies and knowledge to serve the training and teaching
activities. Teaching and researching in research institutes, Academies,
Schools in the Army.
7. The layout of the dissertation
The whole thesis consists of 128 pages presented in 4 chapters with
the Introduction, Conclusion, List of published scientific works,
References and Appendixes.
Chapter 1. OVERVIEW OF HIGH MEASUREMENT METHODS
AND PROCESSING OF SIGNALS OF HIGH CHANNEL OF
CRUISE MISSILE
1.1. Overview of high measurement methods
On general flight vehicles, to measure the altitude of a flying device
typically uses two measurement systems: The high measurement system
does not use magnetic (no-radio) electromagnetic waves and high
measurement systems using electromagnetic (radio) waves.

1.2. The situation of research on processing and combining high
measuring signals
1.2.1. Research situation in the world
The application of SOA for the construction of the model used to
extrapolate the error according to the horizontal channel INS [48], [53],


3

[54] yet clearly mentioned explicitlyin the form of algorithms, evaluation
results also improved qualitative characteristics, especially not the
proposal was time of application algorithms. The application of SOA for
combined processing problems in high measuring combination when
changing flight conditions has not been resolved.
1.2.2. Research situation in our country
The research published in the country on the construction of
algorithms to combine high measurement signals is performed by
Kalman filtering algorithm and given certain results on improving the
quality of processing and combining high measuring signals. . However,
these studies have not fully addressed the limitations and remedies that
result in high combined measurement errors when using Kalman filters.
From the analysis of domestic and foreign research situation, the
thesis raises the problem to be solved.
- For high-measurement combinations with many high-altitude
measurement units combined with different flight conditions, the
algorithm-based optimization selects a high-order measurement structure
combined in a high-measurement combination in one thing. Specific
flight packages are needed.
- In the case of the observation of improper state variables, the
Kalman filter works ineffectively for a certain period of time, using a

model-building algorithm to extrapolate an alternative state error,
resulting in Adjusting the error of INS state.
1.3. The problem of combining high measurement signals on cruise
missiles
When processing signals in a combined high-level measurement,
compensation or correction methods are often used. The synthetic results
are treated with various filters such as Kalman filter, adaptive filter ...
after the filter receives the error estimates of the high-altitude signal
treated with markedly improved quality in the flight of the missiles
1.4. Application of Kalman filtering algorithm and the problem of
selecting structure in high measurement combination
1.4.1. The Kalman filter algorithm treats the combination of high
measurement signals


4

1.4.2. Select a high-gauge set structure in combination with the
observation standard
In the combination of high measurement the high combined meters
are measuring a generic parameter is the elevation, to consider the
remaining components of the state vector to assess whether to calculate
(the observed) and calculate exactly how (the level observed) through the
component is measured directly as a height.
Selection, search out the combined height structure consistent with the
specific flight conditions in order to improve the quality of the processed
signal height combinations when the combination of high measurement
are more combined height, use standard reviews the level of observed
status variables. Interms of specific flights, the quality of the measuring
signal match processors willbe increased markedly, when the combined

height level of the larger states respectively, will choosing to handle the
high measuring signal matching.
1.5. The instability of Kalman filtering algorithm and application
of self-organization algorithm to improve processing quality
combined with high measurement signal
1.5.1. The instability of Kalman filtering algorithm and method of
constructing extrapolation model
In order to overcome the decomposition of the Kalman filter, many m
ethods are given as the offset method, the Kalman filter structure method,
the method of constructing the Kalman filter structure is more
appropriate, however, when initial prognosis is incorrect. (mathematical
modeling, input interference, measuring noise etc.), the use of the above
methods does not bring high efficiency. Alternative methods can then be
used instead: neural networks, self-organizing algorithms, genetic
algorithms [54], [63], [64]…
That at some time (tAtB) to use the model construction algorithms
extrapolate (figure 1.8) then get estimates before time tA the set sample
value zi = z1, z2, z3, …,zN updated results matching measuring signal
processing of high, the algorithm will use this data to evaluate new
construction extrapolation models from the base models


5

Figure 1.8. Overview of constructing models of extrapolating:
BH- The base height; MCA- Model construction algorithms; PA- Prediction algorithm.

1.5.2. Self-organizing algorithm in processing high-signal matching
When the high-level detectors combine signal processing using Kalman
filter, it is not allowed to evaluate the system state accurately enough

with high measuring intensity. Then, the criteria for assessing the
observed level of state variables for the evaluation value do not exceed
the observation threshold, the signal processing results with large errors
for the measurement. At this point, correcting the high measurement
errors of the base measurement system needs to use a new algorithm to
model the extrapolation of their status errors instead. The thesis proposes
to use SOA to solve this problem.
1.6. Conclusion chapter 1
On the basis of analyzing domestic and foreign research works related
to the problem of combining high measurement signals, given the
limitations of the Kalman filter algorithm associated with the combined
high gauge structure, it shows that it is necessary to solve the problem
using the observation level evaluation criteria to optimize the structure in
the high measuring assembly (selection). Choose the appropriate
combination plan - chapter 2). Application of self-organizing algorithm
to model extrapolation of state errors when Kalman filtering algorithm is
not stable (chapter 3).
Chapter 2. ALGORITHMS FOR INFORMATION
PROCCESSING AND SELECTION OF COMBINED HIGH
MEASUREMENT SET STRUCTURE
2.1. Diagram of structure adjustment parameter of high channel
state


6

When connecting the circuit of feedback to the corresponding point in
the structure IHM, Kalman form status equation the following discrete
[56]:
(2.1)

xk  k,k 1xk 1  uk 1   k 1w k 1 ;
k,k-1 is the matrix system; k,k-1 is the input interference matrix;
elements in the matrix k,k-1, k,k-1 received from the mathematical model
of the associated high input signal error; wk-1 is the input noise vector; uk1 is a vector of calibration signals, constructed from optimal estimates of
the filter [56].

accelerometer

- +
+

+ -

TP1

-

-

+
+

+

TP2

+

+

2g/R

g

IHM

Figure 2.1. The scheme of IHM with the feedbacks of the
cumulative error

2.2. Build kinematic model of high measurement error
2.2.1. Differential kinematic model on high channel of inertial
navigation system
The model error of the IHM will be described by differential
equations simpler than [56]:

 H  V

 g
 V   2  H  a  g
 R

 a  a  u
a


 g  g  u g

(2.6)

In formula (2.6) height error H(t) include: the wrong number of

accelerometers under the vertical channel δa(t) = δay(t) and the error due
to the attractive uncertainty g(t)
2.2.2. Kinetic model of error of the radio height meter
The process error Hvt(t) be performed as standard Mackop process


7

satisfy the differential equation a [56]:
 H vt  t   vt  vt  t   u vt  t 

(2.8)

in that:  vt = 1/vt; vt - the constant correlation movement. u Hvt -the white
junk Gauss with mathematical expectation of zero and the correlation
function:





BuH    m u Hvt  t  u Hvt  t     2vt 2vt   t   
vt

(2.9)

2.2.3. Kinetic model of error of the micro-barometer
The cause of the error is primarily the impact of the noise sourcedue
to the fluctuations of motion speed cruise missiles, satisfy the quadratic
linear differential equation for [56].


'
 H ka   H ka
;

 H' ka  (0, 67V   ) H '  0, 67 V  H  3, 2.10 4  u
B
ka
ka
ka B
ka
ka ka


(2.12)

in that: ka- Constant pressure differential correlation, VB- High-speed
change; uka - The speed of the flying device; uka -the white junk input
noise described in the form of random process stops with the correlation
function.
 . 
(2.13)
BH     2H e
ka

ka

ka

2.4. Develop standards for assessing the observed level for combined

high measuring sets
2.4.1. Observed and controlled according to Kalman standards
On the level observed by В.Н. Афанасьев and К.А. Неусыпин there
is a specific concept, considered the precision of approximation of the
status vectors and analyzed the measurement noise as well: the
observability level defines the variance ratio of an arbitrary status
element and the variance of the status vector is measured directly
considering the variance of the measurement noise
2.4.2. Develop standards for assessing the level of observation in the
high-altitude meter IHM-RHM
When considering the exact
characteristics of the high measuring
IHM-RHM need to transfer the persistent Kalman filtering algorithm to
discrete, we have the equation and state equations in discrete form for


8

measuring the combined height IHM-RHM (2.19), (2.20).

in that:

xk  k,k 1xk 1   k,k 1w k 1

(2.19)

z k  H k xk  v k

(2.20)

 H 
 1
 V 
 2gT / R

;



x k  a

 E  F(t k )T   0

 k,k 1

 g 
 0
H vt 
 0




0 1  T
0
0 

0
0
1  T

0 
0
0
0
1  vt T 

T
1

0
T

0
T

0
0

;

0
0
0



;
2
2
T

T
0


T

0
0
 k  k,k 1 (t k )T    T  T  1


0

T
T


1
0





0
0
T  Tvt  1 


Initial conditions: x(0 / 0)  0 ; P  0 / 0  diag 2vt 2v 2a 2g 2vt  ;



Equations in matrix form measurement.
z  x k   O*z*k

(2.24)

Suppose that when calculating the level of observed status vector
components of the heating system only includes ameasurement, that is in
the case Hk = [1 0 ….. -1]. Measurement of equation z(xk) in scalar
form, with the size of the matrix system of n = 5 would be:
z  H   z k  z k  4
(2.25)
z  V    21,k z k   22,k z k 1  .....   25,k z k  4
z  a   31,k z k  32,k z k 1  .....  35,k z k  4
z  g    41,k z k   42,k z k 1  .....   45,k z k  4

The coefficient of i j ,k ( j  1, 2, ..., 5)
is the row of a matrix O* at time
tk. Calculate the variance of the error estimate of the state variables
at time k according to the formula (2.26).
n

M  x i,k   


2

 xˆ
k 1

n

2
i,k

(2.26)

in that: i = 1,2,3,4 corresponds to the element status H,V,a,g.
For arbitrary elements of the statusvector, vector measurement of


9
*
* *
derivatives k  O vk in scalar form corresponding (2.27):

*ik  i1,k vk  i2,k vk1  .....  i5,k vk4

(2.27)

in that:  is the ith element of the vector *k
*i
k

The variance of the ith element measured impurities (R*i) can be
identified by the number system ij, k ( j  1, 2,...,5) will is:
2
2
2

R *ik   i1,k    i2,k   ....   i5,k   R 0k



(2.28)

in that: R 0k is the variance of the original measured vk.
The expression on standards to rate the status element:
D  x i,k  

2
M  x i,k  


2
M  H     ij,k 

 j1
5

(2.30)
2

in that: M  x i,k 2  is the variance of the ith element status vector;




M  H   variance is the status vector measure directly.



2

From (2.30) which in turn determines the level of the state variables
of the IHM. The expression evaluates the level ofobserved status vector
component of velocity (2.31), of the acceleration (2.32), of gravity (2.33)
- Standard construction the criterion for evaluating the observability level
in the high measure of IHM-AHM according to the measurement
equation (2.35)
- Standard construction the criterion for evaluating the observability
level in the high measure of IHM-RHM-AHM according to the
measurement equation (2.37)
2.5. The algorithm selects a high measurement set structure in
combination with the use of a standard of observation
The basic steps of the algorithm choiceusing standard structure
the
criterion for evaluating the observability level :
- Form measurements from the combined height (2.25),(2.35), (2.37)
- The variance of the error estimate of the status vector element in the
set of matching height:


10
n

M j  x i,k   


2

 xˆ
k 1

2
i,k

;

(2.38)

n

in that: i = 1,2,3,4 corresponding to the state variables of the ĐCQT
H,V,a,g; j = 1, 2, 3 is the index correspond to the sets of measures
combining IHM-RHM, IHM-AHM, IHM-RHM-AHM
- Magazine variance measure corresponding to each element in the
status vector measure combines high:
2
2
2
(2.39)
R *ik   i1,k    i2,k   ....   im,k   R 0k




in that: i = 2,3,4 corresponding status vector elements V,a,g; m =
5,6,7- the size of system matrix correspond to the sets of measures
combning IHM-RHM, IHM-AHM, IHM-RHM-AHM; k- the time
calculated at time tk

- The level of the state variables in the set of matching height.
Choosing the appropriate height structure on the basis of reviews the
level of observed status variables.
M j  x i,k  


D j  x i,k  
2
m
2
M  H    j1  ij,k 


2

(2.40)

in that: i = 2,3,4 respectively with the status vector element V,a,g; m
= 5,6,7- the size of system matrix correspond to the sets of measures
combining IHM- RHM, IHM- AHM, IHM- RHM-AHM; j = 1,2,3- the
index correspond to the sets of measures combining IHM-RHM, IHMAHM, IHM-RHM-AHM; k the time calculated at time tk
2.6. The conclusions chapter 2
In this chapter, the error model of a number of high measuring sets
has been developed, constructing the structure of the high-measurement
assembly to select the combined high-gauge structure. Research and
standard analysis assess the level of observation of state variables in
combined information processing.
Applying the research results, developing algorithms to select the
appropriate combination of high gauges by assessing the degree of
observation of state variables to improve the quality of processing high

measuring signals for the selected combination of high gauges when
considering characteristic parameters corresponding to specific flight


11

conditions.
Proposing measures to overcome the limitations of the algorithm in
case the ability to observe state variables does not exceed the observed
threshold, the initial a priori information is not sufficient for Kalman
filtering algorithm.
Research results of chapter 2 will be proved by simulation in chapter
4 and shown on works [3], [5] of the author.
Chapter 3. BUILDING EXTRAPOLATING MODEL IN HANDLING
HIGH-QUALITY SIGNALS ALGORITHMS APPLICATION SELFORGANIZATION

3.1. Basic principles when implementing SOA
3.2. The structure of the SOA
The construction of self-organizing algorithm consists of the
following basic steps:
3.2.1. Enter input data base
Fact based on the experimental process of research subjects, the data
about the technical characteristics of measuring complex high on a
particular cruise missile classand level, the capacity estimated by the
design of variable trend status error will choose the basis functions and
limits the number of basis functions properly and considering it is the
wrong model of simplified base accepted. General form of the basis
functions arespecified in section 1.5.2.1;
3.2.2. Organize improving the quality of the model
The method of finding the coefficient for the linear model [67].

Data is divided nto 2 parts: A- school section; B- the test section is
described on figure 3.2. To time tm in the study A will for a value y(tm).

Figure 3.2. Performing split model to construct and model reviews

To find the coefficients for the linear format model, using the method


12

of minimum squared in A test set [67].
With combinations of y(a,t) depends on the linear coefficient vector
A: y  a, t   a1g1  t   a 2g 2  t   a 3g 3 t   .....  a ng n t 
A   GTG 

in that:

1

G Y 
T

(3.8)

*

 y  t1  
 g1  t1  g 2  t1 




yt 
g  t  g2  t 2 
Y*   2  ;[G]   1 2
 ..... 
 ....
....



y
t
g
t
g




tm 
m
1
m
2




.... g n  t1  
 a1 


a 
.... g n  t 2  
;A   2 

 ... 
....
....

 
.... g n  t m  
a n 

;

To enhance the complexity of the model can use the methods of the
following organizations:
- Organized by combination method: Each level of complexity i of the
model is the short-i of n complex models, the model is a combination of
the form of the selected combination function;
- Organized by selective combination method: Each level of
organization will pick out a fixed amount the best models, the best model
that will participate in the next organization level and aggregated with all
of the remaining base model;
- Organized by the method of multi-sequence iteration: Each level of
organization will be fixed the number of the best model was selected
(each model is considered as a variable). At times the next organization
each pair of variables is taken at random from among the best models
that will the combination out of a new model in the form of fixed
combination function;

- Organized by the repeat method software: Similar to that held by the
method of multi-sequence iteration, this organization methods each
organization level select fixed number of best model. In the next level of
organization, with other organizations under the organization method of
multi-sequence iteration, the best model was selected at the organization
level before will the combination with the original base model remains
tonew models out combinations according to the functional form of fixed
combinations.
3.2.3. Evaluate the selection of models
After each level the organization of certain organization methods,
conducting reviews and select the optimal model to select best


13

extrapolating model or select the number of the best model to use for the
next organization level. To evaluate the new constructing extrapolating
model after each level of organization, we use the standard
assessment in item 1.5.2.2 on the bases of value patterns in previous
Kalman estimation process.
However, the need to incorporate the criteria for evaluation and
selection of models extrapolate according to form:
(3.9)
  w1n dc2  w 22  B  w 3Bi2 ;
trong đó: n dc2 ; 2  B ; Bi2 is the standard minimum shift; uniform standards
and criteria of balance is calculated by (1.27),(1.23), (1.34);
w1, w2, w3 is the weights (w1 + w2 + w3 = 1), the value of wi
depend on reviews of the designer about the importance of each criteria,
if the standard would be considered more important than the value of the
corresponding wi will have greater value than the remaining weight.

When execute the algorithm self -organization, it is often combine
standard shift state  n dc2  and uniform standards  2  B to evaluate and
select extrapolating model, the evaluation and selection of extrapolating
model not only between new constructing models at each organization
level, but also to proceed to select the best model between pattern of
organization level.
3.2.4. Conditions for the end of the algorithm
Based on the value evaluation model extrapolate according to (3.9);
self-organization algorithm will stop when satisfied one of the following
two conditions:
- If the value of reviews () of the best models of organization level
are considering the larger the value of reviews () of the best models of
organization level, the algorithm stops;
Means: level (k+1)>level k; k = 2, 3, 4, ….,n; n- the number of the base
model in the input data of the algorithm;
- With the various problem conditions will advance to the algorithm
stops after some level of organization or after a working time of the
algorithm. This weighing on demand computer fast impact of math, as
well as the complexity of the model and the possibility of dedicated


14

computer calculations, the designer will put time self-organization
algorithm for matching. The result of the best model would be valuable
model sreviews the smallest of levels of organization have made of the
algorithm.
From the analys is of the structure of the self-organizing algorithm
with four basic tasks are done; I have the save map of self-organizing
algorithm is shown on figure 3.8.

Figure 3.8. The scheme of self-organizating algorithm of signal processing

3.3. High-altitude combination of SOA applications
In figure 3.9 introduces high measuring complex diagrams using self
organizing algorithm of constructing and extrapolating model selection.

Figure 3.9. Automatic height combination chart using of SOA: H -is the practical
height information which we need to measure; xk - the error vector IHM; 𝑥𝑘 - vector
model extrapolation of errors; EOL - evaluating the observability level.

3.4. Develop an extrapolation model of error of SOA application
state variables


15

3.4.1. Collect and build input base data for SOA
Choose data base algorithm input the wrong number of selforganization of the altitude is simplified by the variability of the base
model 24 (3.10).
3.4.2. Improve the quality of the model by selective organization
method
In the method of selected complex organizations, each organization
level enhance the complexity of the model will pick out a fixed amount
the best models the best models, will participate in the next sequence
and combination with allremaining models.
Select the funtion associated to the organization level are linear:
(3.11)
y  a,t   y  g1,g 2 ,....,g n   a1g1  t   a 2g 2  t   ...  a ng n  t  ;
The standard of review for the models built for the level held:

(3.14)
3.4.3. Check the stop condition of the algorithm
After each organization level reviews to select the best model. Using
standard algorithm reviews by (3.14), while the value increased rating,
meaning the good level of the model decreased then the algorithm would
stop.
The results of calculations for the advanced degree organization level
of complex models when constructing the model error of extrapolating a
height ofitem 3.4.2:
y(10) (a, t)  a1(10) t1,5  a (10)
t 0,5  a 3(10) t  a (10)
sin(0,1.t)  a 5(10)sin(0,2.t) 
;
(3.19)
2
4
a (10)
sin(0,5.t)  a (10)
sin(t)  a 8(10)exp(5.t)  a 9(10) .exp(3t)  a10(10)exp( t)
6
7
a1(10)  0.574992279351559;a (10)
 0.171280269751940;a 3(10)  0.00784684354414533;
2
a (10)
 0.0212227452780124;a 5(10)  0.0446385963644056;a 6(10)  0.0181295025740750;
4
a (10)
 0.00736648717194733;a 8(10)  0.245347972257312;a 9(10)  0.215674527823339;
7

a10(10)  0.288322982796649.

The algorithm can be stopped in the following terms the number of
predetermined organization level, by the designer based on the data and
computing power of dedicated computer.


16

3.5. Construct the model error of extrapolating the velocity and
acceleration of self-organizing algorithm application
The input data to construct models that extrapolate the velocity error
and the error under acceleration (3.20), (3.21). Construction and
evaluation of model error in the process of improving the quality of the
model.
3.6. The conclusions chapter 3
In this chapter, we present the structure of self-organizing algorithm
to model the extrapolation of errors of state vector elements and apply
them in case of the observable level of the state vectors below.
Threshold, the Kalman filter is diverged during processing associated
with a high measuring signal.
Implement self-organizing algorithm to model extrapolation of height
error of state vectors by selective combination method when organizing
to improve the quality of the model. Assess the advantages and
disadvantages when selecting organizational methods to improve the
quality of the model in the process of building extrapolation model of
self-organizing algorithm.
It has been demonstrated by the algorithm itself to build an
extrapolation model to correct the status error of the high combined
measurement when the Kalman filter algorithm is not stable (divergent

Kalman filter).
The research results will be proved by simulation in the following
chapter and shown on the author's works [1], [2], [5].
Chapter 4. DESCRIPTION OF SURVEY, ASSESSMENT OF
HIGH-QUALITY SIGNAL HANDLING
4.1. Set simulation with assumptions, input data of the problem
4.1.1. Building simulation program
Construction of simulation algorithm of program reviews the high
measurement quality improvement associate use standard reviews the
level of observed and self-organization algorithm construct extrapolate
errors. The steps are done according to the save map image algorithm
figure 4.1.


17

Figure 4.1. The scheme of the algorithm to enhance the quality of
processing measurement signals higher combined standard the criterion for
evaluating the observability level applications and self-organization algorithm

4.1.2. Assumptions, data for simulation
- In the dissertation will not go to analyze the dependence of
characteristic parameters of the object of research on the fluctuations of
atmospheric gas and earth to choose the correct characteristic parameters
with the fact. That the value of this characteristic paramet; terms of
assumptions and retrieved the document [56];
- The steps preliminary to the signal processing board for the
parameters from the measured high and the conversion between the
coordinate system as was done. The causes of the error from the high
measurement related to hard ware calibration before was put to work;

- The orbit plane of a dissertation review class cruise missiles with
altitude strip mixture;


18
- The choice of input data for self-organizing algorithm base on the
rules change the error of the status element on the basis of the data
set to assume (3.10), (3.20), (3.21).
4.2. Survey results of algorithm for selecting combined high gauge
structure

0,435

0,293

0,162

Figure 4.2. Evaluating the observation level
of the velocity error of combination height
measurement at vt =10s, ka =25s,
g =9,7803 m/s2

Figure 4.3. The height error by using the
Kalman filter of combination height
measurements when vt =10s, ka =25s,
g =9,7803 m/s2

In figures 4.2, 4.4: 1- The observability level δV of the complex IHMRHM; 2-The observability level δV of the complex IHM-AHM; 3- The
observability level δV of the complex IHM-RHM-AHM.

0,325
0,248

0,094

Figure 4.4. Evaluating the observation level
of the velocity error of combination height
measurement when vt = 15s, ka = 30s,
g = 9,7786 m/s2

Figure 4.5. The height error by using the
Kalman filter of combination height
measurements when vt = 15s, ka = 30s,
g = 9,7786 m/s2

In figures 4.3, 4.5: 1- Evaluating the height error of the complex
IHM-RHM; 2- Evaluating the height error of the complex IHM-AHM; 3Evaluating the height error of the complex IHM-RHM-AHM, 4- The
actual height error;


19

Bảng 4.1. Evaluating the height error of combination
height measurements
The mean of the error
(m)
IHM-RHM
IHM-AHM
IHM-RHM-AHM

H =15m
0,0205
0,0254
0,0293

The variance of the
Standard deviation (m)
error (m2)

H =14km
0,0427
0,0221
0,1986

H =15m
0,000325
0,000532
0,000686

H =14km
0,00126
0,000373
0,00242

H =15m
0,018
0,0231
0,0262

H =14km

0,0355
0,0193
0,162

4.3. Application of the standard for evaluating the degree of
observation with improved processing quality for the high IHM-HM

0.268
0.175

Figure 4.6. The observability level of the
velocity error in different discrete
intervals T =0,1s (graph 1) và T =0,2s
(graph 2)

Figure 4.7. The velocity error by using the
Kalman filter at different T: 1- The actual error
value; 2- The error value with T = 0,1s; 3- The
error value with T = 0,2s.

Bảng 4.2. Evaluating the velocity error at different T in the set status
Discrete
range (s)

The mean of the
error (m/s)

The variance of
the error (m2/s2)

Standard deviation
(m/s)

0,1

1,241

1,994

1,412

0,2

2,514

7,518

2,742

Figure 4.8. The observability level of the
velocity error when vt change: 1- The
observability level of the velocity error when
vt=5s; 2- The observability level of the
velocity error when vt=20s

Figure 4.9. The velocity error in combination
processing the high measurement signals by
using the Kalman filter when vt change

20

In figures 4.9: 1- The actual error value; 2- The observability level of
the velocity error when vt=5s; 3- The observability level of the velocity
error when vt=20s.

Figure 4.11. The acceleration error in
combination processing the high measurement
signals by using the Kalman filter when vt
change: 1- The actual error value; 2-The
observability level of the acceleration error
when vt=5s; 3-The observability level of the
acceleration error when vt=20s.

Figure 4.10. The observability level of the
acceleration error when vt change: 1- The
observability level of the acceleration error
when vt=5s; 2- The observability level of
the acceleration error when vt=20s.

Bảng 4.3. Evaluating the velocity error, the acceleration error
at vt different.
The mean of the error
(m/s), (m/s2)
velocity error
acceleration error

The variance of the error Standard deviation
(m2/s2), (m2/s4)
(m/s), (m/s2)

vt =5s

vt =20s

vt =5s

vt =20s

vt =5s

0,8352

1,3565

0,5964

1,8692

0,7723

0,568.10

-2

0,973.10

-2

0,1332. 10

-4

0,6839.10

-4

vt =20s
1,3672
-2

0,365.10 0,827.10-2

4.4. Assessing the quality of processing and combining high
measuring signals using self-organizing algorithm

Figure 4.12. The velocity error in combination processing the high measurement
signals by using the Kalman filter and using self organizing algorithm

In figure 4.12, 4.13: 1- The actual error value. 2- Evaluating the error
by using the Kalman filter; 3- Evaluating the error by using self
organizing algorithm


21

Figure 4.13. The height error in combination processing the high
measurement signals by using the Kalman filter and using self organizing
algorithm
Table 4.4. The table of values

of evaluating the velocity error
Time (s)

The mean of the error (m/s) The variance of the error (m2/s2) Standard deviation (m/s)

40  200
200  280

Kalman

self organizion

Kalman

1,2749

2,1536

0,9332

17,962

2,985

3,0848.10

self organizion Kalman self organizion
2

5,626

0,966

2,372

8,1396

17,5638

2,853

Table 4.6. The table of values of evaluating the height error
Time (s)

The mean of the error (m) The variance of the error(m2) Standard deviation (m)
Kalman self organizion

Kalman

40  200

0,0244

0,0896

3,2446.10

200  280

0,3997

0,0892

0,6391

self organizion Kalman self organizion
-4

0,0045

0,018

0,0670

0,0297

0,7994

0,1723

4.5. The results of the algorithm survey organize the construction
and select extrapolation models

Figure 4.14. The height error of combination Figure 4.15. The height error of combination
height measurements at H =15m when
height measurements at H =14km when

In figures 4.14, 4.15: 1,2,3- extrapolate using self organizing
algorithm; 4,5,6- using the Kalman filter; 7- Evaluating of the actual
error value; (in that: 1,4- the height error value of IHM-RHM; 2,5- the

22

height error value of IHM-AHM; 3,6- the height error value of IHMRHM-AHM).
Table 4.7. Evaluating the height error of combination
height measurements
Time (s)

IHM-RHM
Kalman self organizion

IHM-AHM

IHM-RHM-AHM

Kalman self organizion Kalman

self organizion

The variance of the height error (m2) at H =15m
t =(40 ÷200) 0,000325
t = (200 ÷280)

0,6391

0,0045

0,000532

0,0081

0,000686

0,0116

0,0297

0,8577

0,0358

0,9789

0,0527

2

The variance of the height error (m ) at H =14km
t = (40200)

0,00126

0,0099

0,000373

0,0060

0,00242

0,0452

t = (200280)

0,9671

0,088

0,6629

0,0698

0,9944

0,1357

4.6. Conclusion chapter 4
Chapter 4 has conducted simulation evaluation of processing quality
combined with high measuring signals, when using the observation level
assessment criteria to optimize the structure in the high measuring
assembly, select the set structure. Combined high gauge suits specific
flight conditions. In case the observation level of state variables is
smaller than the observed threshold, the Kalman filter or more diverges,
using the SOA to model the state error extrapolation model instead.
The simulation results confirm that the processing accuracy associated
with high measuring signals is increased when:
- Using the criteria for evaluating the degree of observation of the
state variables, selecting the high measuring structure and optimal
combination in the high measuring combination.

- Using self-organizing algorithm to model extrapolation of state
errors in processing in combination with high measuring signals.
CONCLUSIONS
The thesis has studied and applied modern algorithms and evaluation
criteria to combine high measurement signals to optimize high
measurement, improve the quality of control process - stabilize the height


23

of cruise missiles when conditions change. Research results of the thesis
have built intelligent high-altitude combination using standards to assess
the level of observation and self-organizing algorithms in combining
high measurement signals
The thesis got main results as follows
1. We analyzed clearly the eveluation criterion (of the quantity) of the
observability degree and the self-organizing algorithm for constructing
an extrapolation model. Applied them in the combination process of
height measurement signals
2. We improved the quality of the combination process of height
measurement by choosing a suitable structure of the combined height
meter in a height measurement combination with many working height
meters based on use of the valuation criterion of the observability degree
of the status vector's components among combined height meterd within
the same flight conditions
3. We improved the quality of the process of height measurement
signals for the combined height meter when considering the
characteristic parameter of the combined height meter corresponding a
concrete flight condition. The obtained results showed that when the
observability degree of status variables increases, the exactness of the

combination process of height measurement signals increases
4. When the status variable has a low observability degree due to the
action of external conditions, ues of the self-organizing algorithm for
constructing an extrapolation model of the status error auguments the
exactness of the combined process of height measurement signals.
5. The research results were tested by solfware Malab/Simulink on
computers and verified the correctness of proposed algorithms.
Summary of new conclusions of the dissertation
1. Construction of an algorithm for the combined process of height
measurement signals and choice of the structure of the combined height
meter based on the evaluation of the observably level.
2. Construction of an extrapolation model in the combined process of