A review of FFT algorithms and a real-time algorithm development for airborne vibration testing applications

Abstract

Data collection is a critical aspect of flight test instrumentation as it enables the evaluation of aircraft performance and safety through the monitoring of critical parameters such as vibrations, temperature, pressure, and force, as well as digital bus operations. However, effective communication strategies are necessary to transmit this data, particularly high-frequency vibration data, during prototype testing of full-scale air vehicles. The Fourier Transform is a fundamental technique that decomposes signals into their constituent frequencies, thereby facilitating the analysis of specific components within continuous data. The primary objective of this study is to conduct a comprehensive analysis and assessment of multiple Fast Fourier Transform (FFT) algorithms to determine the optimal approach for addressing practical aviation challenges. The selection of the most appropriate FFT algorithm can significantly influence the efficiency and precision of data analysis in the context of flight test instrumentation. The case study offers comprehensive insights into the utilization of the chosen FFT technique to handle a specific difficulty related to aviation. This indicates the practical importance of this research in the analysis of real-time vibration measurements. This study introduces an algorithm that has been extensively developed for the purpose of analysing real-time vibration data. This algorithm provides a comprehensive understanding of the many features of signals within the frequency domain. The technique highlights the Chirp-Z Transform (CZT), which facilitates the accurate finding of frequency elements. The selection of the Hann window in the CZT analysis is crucial for mitigating spectral leakage, hence enhancing the precision of frequency analysis. During the spectrum resolution and pre-processing phase, the vibration data is carefully divided into overlapping windows, with parameters computed to improve frequency resolution. By applying oversampling with a 512-point Hann window at a sampling rate of 2048 Hz, the resolution is enhanced, enabling a more comprehensive analysis of the frequency spectrum. The increased level of accuracy demonstrated by this measurement technique serves to validate its importance in the detection and analysis of high-frequency harmonics. Thus, it facilitates a greater understanding of engine functionality and provides possibilities for potential progress in this domain. The technique of collecting vibration data employs the usage of the PCB 33931 vibration sensor, which effectively translates mechanical vibrations into corresponding electrical impulses. To enhance precision, the raw data undergoes signal conditioning utilizing Curtiss-Wright hardware. This process involves noise reduction, amplification, and filtering. The efficiency of the algorithm is achieved by optimising computations across numerous CPU cores to minimize processing time. The real-time simulation of CZT is executed on a personal computer equipped with a 12th-generation Intel Core processor, particularly the i7-12700H model, operating at a frequency of 2.30 GHz. In an extensive assessment including the central processing unit (CPU), hardware components, and processor, the mean duration per iteration, which contains the process of data collecting, is 0.0315 seconds. The comprehensive analysis offers an overall view of the computing process, considering multiple elements that influence performance. However, after conducting a more detailed examination, with a specific emphasis on the CZT analysis, it becomes obvious that there is a significant increase in efficiency. When isolating this one aspect from the whole algorithm, the time required for each iteration exhibits a considerable decrease to only 0.0017 seconds, indicating a significant boost in efficiency.