FUDGE FACTOR/ACCURACY DETERMINATION

The uncertainty bounds for parameter estimates obtained from the Cramer-Rao bounds/OEM are multiplied with a fudge factor to reflect the uncertainty correctly. This is because when the OEM (which does not handle the process noise) is used on flight data, which are often affected by the process noise (like atmospheric turbulence), the uncertainty bounds do not correctly reflect the effect of this noise on the uncertainty of the estimates. And hence a fudge factor of about 3 to 5 can be used in practice.

TABLE C2 Design Approaches and Performance Specifications
Bode Plots/Frequency Responses	Nichols Charts/Frequency Responses
Time-Domain/Optimala Pole Placement Method Control Methods
Root Locus
Performance criteria to be specified	Coefficients of steady-state errors/phase margins/crossover frequency	Coefficients of steady-state errors/maximum closed loop frequency response and frequency	Coefficients of steady-state errors/location of dominant poles of the closed loop and root sensitivity Cascade lead, lag, feedback	Desired closed loop TF, poles sensitivities to parameter variations More general ones in feed forward, feedback, or both	Integral performance criteria in time domain
Compensation Cascade lead, lag, or types lead-lag	Cascade lead, lag, or lead-lag
Varieties of optimization methods are used for obtaining the optimum input to the plant
Source: Sinha, N. K. Control Systems. Holt, Rinehart and Winston, Inc. 1988. a There is a huge body of literature on optimal/modern control methods. Besides these, the approaches based on quantitative feedback theory, H-infinity, and hybrid (H2/H-infinity) methods are also used.

This number has been arrived at (on the average) by performing flight simulations – based parameter-estimation exercises for a fighter aircraft for longitudinal and

lateral-directional maneuvers and using the formula FF = Jn sa™plinf frequenC^—.

° V (2 bandwidth of residuals)

C10 GENETIC ALGORITHMS

Genetic algorithms are heuristic/directed search methods and computational models of adaptation and evolution based on the natural selection strategy of evolution of biological species. The search for beneficial adaptations to a continually changing environment in nature (i. e., evolution) is fostered by the cumulative evolutionary knowledge that each species possesses of its forebears. This knowledge, which is encoded in the chromosomes of each member of a species, is passed from one generation to the next by a well-known mating process wherein the chromosomes of ‘‘parents’’ produce ‘‘offspring.’’ Thus, genetic algorithms mimic and exploit the genetic dynamics underlying natural evolution to search for optimal and global solutions of general combinatorial optimization problems. The applications are the traveling salesman problem, VLSI circuit layout, gas pipeline control, the parametric design of an aircraft, learning in neural nets, models of international security, strategy formulation, and parameter estimation.

[1] dL

Ix dp

[3] Using a 4DOF lateral-directional model, find the eigenvalues for spiral, roll, and DR mode, and (3) determine the roots using DR approximation. Compare the results with those obtained in (2) and comment.

5.6 How would you get the flight path rate from acceleration and subsequently the body rate?

5.7 Will large or small changes occur in the airspeed in the SP transient mode?

5.8 What is the main distinction between SP and phugoid mode characteristic from the motion of the aircraft and the interplay of the forces and moments?