Analytical study for the methods using voice for determining the radial velocity of the airplane.

number: 
525
English
department: 
Degree: 
Imprint: 
Physics
Author: 
Ahmed Shakir Mahmood Yas Al-Hyali
Supervisor: 
Dr. Loay Adward Jorj.
Dr.Laith Abdul-Aziz Al-Ani
year: 
2001
Abstract:

The sound of the moving airplane consists of wide range of drifted frequencies. So to estimate the behavior of the change of some characteristic frequencies the sound wave form_data was chopped into short time segments, Each segment was passed through a Fourier transform to extract the values of the characteristic frequencies. An Adaptive peak detection algorithm was designed and applied on the power spectrum. This algorithm utilizes two types of thresholds in order to avoid the detection of very low frequency peaks. This algorithm can be applied for more than one time in order to detect the major peaks distributed along the interval [fmin, fntax] of the power spectrum. In the current research work a novel method called Histogram Summation Method was designed to determine the characteristics frequencies. This method is based,] on the concept of stability for the characteristic frequencies . The degree of the characteristic frequency ability was considered as "'the percentage the occurrence of the frequency within certain number of successive frames". Two tracking methods for the shifted frequencies were proposed and applied, these methods are Free Tracking Method and Distance Tracking Methods. The first method was based on performing a bi-directional scan of moving window. While in the 2nd method two control factors are taken into consideration, the direction of moving window and the relation between frequency shift due to Doppler effect and the characteristics frequencies. In order to improve the handling procedure for the case of missing frequency an adaptation was performed by increasing the window size for the last shifting frequencies to be greater than the corresponding window size. For simplifying the Doppler shift calculations a simplified theoretical model was considered. Through the analysis of this model it has been shown that frequency variation for short time interval (few seconds) before and after the nearest passing time is linear, and thus the slope of the linear regression between, the frequency and time will represent the rate of frequency change . This rate could be used to estimate the Doppler frequency shift. An algorithm was designed and applied to calculate the frequency shift. In this research work a certain digital sound records for moving airplanes were tested. To take decision about the reliability of the designed system three qualitative measure were calculated, they are: 1.The root mean square error for the relationship between Δf and f. 2.The sign of the slope for the relationship between f and time- which has to be negative. 3. The sign of the slope for linear-regression between Δf and f which should be positive. 4. The dynamic range for the accepted characteristic frequencies. Three parameters have been investigated in our experimental analysis. These parameters are: 1.The frame size. 2. The power spectrum ratio (PSR). 3.The lower frequency threshold (f min) Our results show that the best results for the frame size 0.025 sec, PSR=80% and fmin = 30.