You are on page 1of 21

Numerical Methods in Aerospace Engineering 3 Instrument Landing System Lateral Beam Guidance System

Jekaterina Sazonova 1004263

Variable Notation

Summary

Propeller Lab Report

Page 1

Abstract

Summary

Introduction

The Instrument Landing guidance System is designed to help aircraft pilots align their planes with the center of a landing line during final approach. The ground equipment of the ILS consists of two directional transmitters that send out radio beams from either side of the runway's centerline. The radio pulses are picked up by instruments on the plane and then processed and converted into precise directional and altitude information. The ILS is tied into an automatic pilot, whereby localiser and glideslope transmitters guide the plane into position while those on the aircraft control airspeed by means of an automatic control.

Figure 1: ILS Diagram This is the technical report that outlines the development of the state space model of the ILS and analyses the simulation of the system for the later beam guidance. Using MATLAB and Simulink, the code and the block diagram were generated for the simulation. The aircraft dynamics is then considered and analysed to represent the change in the range of the aircraft. The ILS can be used for automated landing. In order to achieve this direction, descent and speed have to be controlled through automated control systems. The general principle of completely automated landing systems is to feed information from the localiser and glideslope beams into the autopilot of the aircraft. From the initial conditions, using the appropriate mathematical equations, the state space model should be developed and then described through the state variables. The Lateral Beam Guidance System takes a value for the reference angular error between the aircraft and the centre line and compares it with the actual angular error. The comparison of these values is performed by the coupler. The resulting heading angle is then fed into the autopilot to generate an appropriate heading for the aircraft.

Propeller Lab Report

Page 2

Equation Based Simulation of ILS Model

The most important part of the lateral beam guidance system is the lateral autopilot and its interaction with the lateral dynamics of the aircraft. The Lateral Beam Guidance System takes a value for the reference angular error Pref and compares it with the actual error P. The resulting commanded heading angle ]c is then fed into the lateral autopilot:

Where (m) is the lateral displacement and R is the range of the aircraft. and are parameter values found in the Table 1. The Lateral Autopilot has numerous components that provide the actuator commands to the ailerons. When heading angle ]c goes into the autopilot it passes through directional gyro and gains a parameter Kd. Then, it is a representation of the roll angle , which passes through vertical gyro and gains Kv. The output is roll rate p (rad/s):

The roll rate passes through roll gyro and gains Kr. These generate the signal for the aileron deflection (rad), for the banking turn:

Where is the time variable and

is the comparison error:

Where is parameter value. The dynamics of the aircraft are formed through two relationships: the conversion of aileron deflection to roll rate p and the conversion between the roll angle and the heading angle . These form the following two equations:

Where g is the acceleration due to gravity, VT is the velocity of the aircraft and parameter value.

us the

From the equations listed above, the state space model of the system can be developed. First, the state variables should be defined:

Propeller Lab Report

Page 3

The derivatives:

Substitute into the state expression:

In order to calculate the states at each time step we need to integrate the system state derivatives. The algorithms of integration will depend on the step size of the simulation, as will their accuracy. Developing the MATLAB code, which is found in the Appendix, the Euler method is used to integrate the state space model. Euler is a first order numerical integration algorithm which approximates the integral as the first two terms of the Taylors Series.
This is the DERIVATIVE SECTION of the simulation used for the ILS assignment. % % The current time, state and input values are passed to the function as arguments % and the function returns the state derivative. function xdot = system_model(x,psi); global kd kv kr tau Ka Ta g Vt main program % global parameters transferred from

xdot(1,1) = 1/tau*((psi-x(3))*kd*kv-x(4)*kv-x(2)*kr-x(1)); %State Space equation for x(1) xdot(2,1) = 1/Ta*(Ka*x(1)-x(2)); xdot(3,1) = (g/Vt)*x(4); %g is the value of the gravity xdot(4,1) = x(2); xdot(5,1) = x(3)*Vt;

To analyse the system, the variables can be plotted versus time, Figure 2:

Propeller Lab Report

Page 4

Figure 2: Euler Integrated Model Simulation.

From the Figure 2 it can be observed that system follows sinusoidal pattern and oscillation amplitude tends to zero or steady state. Amplitude is defined from the initial conditions. The rate or how fast the system tends to steady state is defined by variable Gc and aircraft flight speed . For comparison, this model can be integrated by other method like Runge-Kutta 4. Fourth Order Runge-Kutta approximates the first 5 terms of the Taylors Series. It does this by making 4 predictions of the state derivatives or slopes of the time interval. These take the form of an evaluation of the slope at the start, 2 predictions of the slope in the middle and then a final prediction of the slope at the end. Because the system does not go to the higher order, the difference in simulation is not noticeable. The Runge-Kutta is more precise method and stores more information, because the number of predictions is higher than in Euler method, therefore it takes longer time to simulate the system. The graphs almost match each other, Figure 3:

Propeller Lab Report

Page 5

Figure 3: The Comparison of Runge-Kutta 4 and Euler Integration Methods.

Simulink Block Diagram

Simulink is an environment for multidomain simulation and Model-Based Design for dynamic and embedded systems. It provides an interactive graphical environment and a customizable set of block libraries that let you design, simulate, implement, and test a variety of time-varying systems. In this section, the state space model was simulated in Simulink, Figure 4:

Figure 4: Simulink System Diagram.

Propeller Lab Report

Page 6

The Lateral Beam Guidance System initially sends the reference Pref =0 to the system. To generate the actual value of P]c gains Gc value in a coupler and then feds it in to the autopilot, Figure 5:

Figure 5: Autopilot. The Autopilot forms 3 loops around the aircraft roll and heading dynamics. The commanded ]c

gains the value of

and

and then passes through the Aileron servos, Figure 6:

Figure 6: Aileron Servo.

The comparison error is then divided by time constant and being integrated to get the value for the aileron deflection , which is then passed to the Roll and Heading Servos:

Figure 7: Roll and Heading Servos.

Propeller Lab Report

Page 7

Where aileron deflection is converted to roll rate p, roll angle and heading angle . Once the heading has been changed so that aircraft approaches the centre line of the runway, it directly influences the lateral speed. This is used to calculate the lateral displacement YR, Figure 4. As YR is divided by range value R, the estimation of the angular error between the aircraft and the centre line can be found. The Simulink should produce same output as in MATLAB and. Therefore, validate the model. The output graphs should match:

Figure 8: Simulink Graphs of the Model Simulation.

Propeller Lab Report

Page 8

The Performance of the ILS System

The performance for the ILS Guidance System depends on many factors. However, the important thing to consider is the actual aircraft heading speed VT and the localiser transmitter

parameters that affects the state variables of the autopilot. This is the section of the report that investigates the system performance based on the aircraft
heading speed VT and compensator parameter value GC , which directly proportional to the command heading of the aircraft. If GC value is increased by approximately 25%, from 48 to 60, it can be observed that the system starts to be unstable. The amplitudes of the oscillations increase, therefore, steady state is impossible to achieve, Figure 9:

Figure 9: The Simulation of the Model, Gc value of 60. Propeller Lab Report Page 9

Theoretically, the increase of the aircraft speed VT should have the same result on the system, as there is an increase in momentum. If the value of the Gc is kept as the initial value of 48 and VT is increased from 60 (m/s) to 75 (m/s), from Figure 10, the continuous sinusoidal graphs with the constant amplitude values can be observed within the 100 s time domain. Therefore, system does not reach steady state:

Figure 10: The Simulation of the Model, VT value of 75 (m/s).

To make the system more stable, these parameters should be decreased. More rapid changes in a system are observed with the VT, as it directly affects the aircraft heading. Decreasing Gc by half and leaving VT with the nominal value, the results are in Figure 11:

Propeller Lab Report

Page 10

Figure 11: The Simulation of the Model, Gc value of 24 and VT value of 60 (m/s). The obvious change can be noticed in the simulation. It approaches zero value much faster, comparing to the initial results. The system reaches the steady state at the time of approximately of 30 s of the simulation.

Propeller Lab Report

Page 11

Figure 12: The Simulation of the Model, Gc value of 12 and VT value of 45 (m/s). As the values of Gc and VT continue to decrease, the simulation become more stable and reaches steady state much faster. From the Figure 12, with Gc value of 12 and VT value of 45 (m/s), the steady state is reached at approximately 30 s, the number of oscillation decreased till 2 and does not go negative, therefore, vector direction does not change for Y R.

Propeller Lab Report

Page 12

Figure 13: The Simulation of the Model, Gc value of 6 and VT value of 30 (m/s). In Figure 13 the simulation becomes even more stable, the pattern is much smoother and the rate of change is decreased. Because if the slower speed, the aircraft heading reaches steady state in the longer time period. The lateral speed decrease becomes more lineal. However, if VT value continues to decrease, the operation of the aircraft becomes impossible. Compering to the previous result, we can observe instability in the aircraft heading in time interval from 10 to 18 seconds. As a result, it is better to leave the Gc value of 6 sustain the VT at 45 (m/s) for better performance.

Effect of the Integral Term in the System

It is normal practice to include an integral term within the coupler of a Lateral Beam Guidance system: Initially, the value of of the system. in the time domain is zero. The change in value affects the behaviour

intlambda=0; lambdaref=0; % Integral intlambda=intlambda+stepsize*((lambdaref-x(5)/R)); % model psi = Gc*((lambdaref-x(5)/R))+intlambda*Gc*ki; xdot = system_model(x,psi);

Propeller Lab Report

Page 13

As the integral value appears in the model, the simulation becomes more unstable. an inverse effect on the model and it becomes harder to reach the steady state.

has

Figure 14: The Simulation of the Model, Gc value of 48 and value of

value of 0.1. ,

From the previous section, the optimal value of Gc of 6 was recognized. The smaller is the better is the performance of the system. should tend to zero for the optimal performance of the system.

Propeller Lab Report

Page 14

Interpolation

So far the longitudinal dynamics of the aircraft have not been considered. To incorporate these dynamics, the range of the aircraft R can be varied. Having the information about the range R change with time and using the appropriate interpolation method, the residuals can be determined. Interpolation is estimation of a value between two known data points. From there, the difference between the numerical value of one particular observation and the theoretical result can be found. The basic idea of interpolation is that there is n pairs of points and we want to find a function P(x) or polynomial that for each xi is equal to yi. The degree of the polynomial depends on the number of points that needs to be interpolated. There are different ways of the interpolation: Newton Interpolation and Divided difference Form, Lagrange Interpolation, Chebyshev-Gauss interpolation and others.

Time(s) 0 10 20 30 40 50 60 70 80 90 100 Range(m) 6000 4673 3639 2834 2207 1719 1339 1043 812 632 4932
Figure 15: Range Data.

Given a set of n+1 data points (xi, yi), i = 0, 1, 2, ., n where the xi are all different and the yi are sampled from an underlying function y = f(x), the nth order interpolating polynomial can be expressed as:

The general approach to interpolate data with Newton Divided Difference Method is based on the evaluation of finite divided differences. Given a set of n+1 data points [xi, f(xi)], i = 0, 1, 2, ., n the finite divided differences of various orders are defined in Equations below:

Propeller Lab Report

Page 15

0 th order 1 st order 2 nd order

f [xi] f [xi , xj,] f [xi , xj, xk]

= = =

f (xi) ,  ,


n th order

f [xn , xn-1, ,x1, x0] =

Note how the divided differences are obtained recursively from two divided differences of order one less. When there are 3 data points (n = 2), the divided differences f [x0 ], f [x1, x0], and f [x2, x1, x0] are identical to the coefficients b0, b1, and b2 and bn.

Figure 16: Table of Finite Divided Differences for a Function f(x).

In general, with n + 1 data points there are n +1 zero order divided differences, n first order divided differences, n-1 second order divided differences, etc. up to one nth order divided difference. The coefficients bi, i = 0, 1, 2,, n of the interpolating polynomial are the first finite divided-difference entries in their respective columns. To calculate the divided differences in MATLAB, two nested loops and general formula is required to create a interpolation function:
function [coef] = newtoninterp(x,y) % Plot the Newton interpolation polynomial P(x) that satisfies P(xi) = yi % coef is the first divided difference for each level x=0:10:100; y=[6000 4673 3639 2834 2207 1719 1339 1043 812 632 493]; nx = length(x); ny = length(y); if nx~=ny, disp('x and y have different length'); return; end % calculate the divided differences dd = zeros(nx,nx); dd(:,1) = y; for i=2:nx % i is the level (the columns) for j=1:nx-i+1 % the divided differences in level i

Propeller Lab Report

Page 16

% Newton Divided Difference general formula: dd(j,i) = ( dd(j,i-1) - dd(j+1,i-1) ) / ( x(j) - x(j+i-1) ); end end coef = dd(1,:); end

Which give us the values for the residuals: Columns 1 through 3 6.000000000000000e+003 -1.327000000000000e+002 1.464999999999999e+000 Columns 4 through 6 -1.066666666666663e-002 5.416666666666551e-005 -8.333333333329854e-008 Columns 7 through 9 -4.166666666667676e-009 1.190476190476453e-010 -2.480158730159310e-012 Columns 10 through 11 4.684744268078674e-014 -8.267195767197460e-016

Figure 17: Interpolated Polynomial.

Propeller Lab Report

Page 17

Conclusion

In conclusion, propellers with high pitch will be more efficient at a lower RPM than a similar propeller with lower pitch. In this experiment the pitch 18 propeller performs the best at lower frequencies (60 to 80Hz), the pitch 15 propeller performs better at the higher frequencies (80 to 100Hz) and it is expected that the pitch 10 propeller would perform the best at even higher frequencies. It was also observed that each propeller performed the best with power consumption in the range of 20 to 24W.

10 Reference List

Propeller Lab Report

Page 18

Appendix

Propeller Lab Report

Page 19

Reference List: Thomson, G. P., 1987. Applied Aerodynamics. Hodder and Stroughton: London. Scanlan, Robert H., 1951. Introduction to the study of aircraft vibration and flutter. New York : Macmillan. Propeller Laboratory Handout. 2010. The University of Glasgow. References for the Figures: Figure 1: Propeller Laboratory Handout. 2010. The University of Glasgow.

Propeller Lab Report

Page 20

You might also like