Author(s)

J. Zhou & Y. Zhong

Abstract

Spectrum-analysis technology has been widely used in engineering. Round discrete spectrum or FFT spectrum of a signal can be obtained rapidly by the FFT method. In practical applications, these FFT spectrums always need to be rectified in order to obtain a better spectrum. A high precision spectrum rectifying technique, a deconvolution-based spectrum rectifying technique, is developed in this paper. The sampled signal is firstly multiplied by an adaptive window function, the zooming spectrum of discrete windowed signal is then calculated, and finally the spectrum of window function is eliminated from the zooming spectrum by deconvolution so that the rectified spectrum of the analyzed signal can be obtained. Experiments show that this rectifying technique is very effective. Keywords: spectrum rectifying technique, deconvolution, window function, sidepetal error. 1 Introduction Frequency spectrum analysis technology has many applications in engineering [1]. FFT is frequently used to obtain signal spectrum rapidly. But the FFT spectrum with low resolution and sidepetal error caused by truncation is too rough to satisfy engineering demands. It induces the study on spectrum rectifying started from 1970s of the 20th century and many achievements have been made [2-4]. One of which is window function spectrum rectifying technique that decreases the sidepetal error by selecting a narrow-band window function. However, so far the rectified spectrum generated by this method is still

Keywords

spectrum rectifying technique, deconvolution, window function, sidepetal error.