An Architecture for On board Frequency Domain Analysis of Launch Vehicle Vibration Signals

Defence science journal

article

R Srinivasan¹  B Lakshmi²  Tessy Thomas³ 

[1]DRDO-Research Centre Imarat

[2]National Institute of Technology

[3]DG

关键词: Modal parameters;    Spectral analysis;    Windowing;    Overlapping;    FFT;    PSD;    Radix-2 BF;    Block RAM;   

DOI  :  10.14429/dsj.73.17690

学科分类：社会科学、人文和艺术（综合）

来源: Defence Scientific Information & Documentation Centre

【 摘 要 】

The dynamic properties of the airborne structures plays a crucial role in the stability of the vehicle duringflight. Modal and spectral behaviour of the structures are simulated and analysed. Ground tests are carried out with environmental conditions close to the flight conditions, with some assumptions. Subsequently, based on the flight telemetered data, the on-board mission algorithm and the auto-pilot filter coefficients are fine tuned. An attempt is made in this paper to design a novel architecture for analysing the modal and spectral random vibration signals on-board the flight vehicle and to identify the dominant frequencies. Based on the analysed results, the mission mode algorithm and the filter coefficients can be fine tuned on-board for better effectiveness in control and providing more stability. Three types of windows viz. Hann, Hamming and Blackman-Harris are configured with a generalised equation using FIR filter structure. The overlapping of the input signal data for better inclusiveness of the real-time data is implemented with BRAM. The domain conversion of the data from time domain to frequency domain is carried out with FFT using Radix-2 BF architecture. The FFT output data are processed for calculating the power spectral density. The dominant frequency is identified using the array search method and Goldschmidt algorithm is utilised for the averaging of the PSDs for better precision. The proposed architecture is synthesised, implemented and tested with both Synthetic and doppler signal of 300 Hz spot frequency padded with Gaussian white noise. The results are highly satisfactory in identifying the spot frequency and generating the PSD array.

【 授权许可】

All Rights reserved