Digitalizer mean filtering method based on density estimation

Abstract

The present invention discloses a digitalizer mean filtering method based on density estimation. The method comprises the steps of: performing oversampling of a digitalizer, performing grouping for anobtained data sample sequence to obtain a plurality of sub samples, adding the sub samples to the density estimation samples, employing a fusion method of density estimation and Bayes information toobtain a probability density function of the data sample sequence, and calculating and obtaining the standard uncertainty of the data sample sequence; and performing re-grouping of the data sample sequence according to the oversampling multiples, obtaining the confidence interval of each sub sample after the repeat groups are obtained according to the standard uncertainty to obtain gross error elimination, and performing average extraction of the obtained samples to obtain final samples. The digitalizer mean filtering method based on density estimation can perform density estimation for the data sample sequence grouping obtained by sampling, and can set the confidence interval based on the density estimation result to perform gross error elimination so as to allow the mean filtering resultto be more accurate.

Description

technical field [0001] The invention belongs to the technical field of digitizers, and more specifically relates to a density estimation-based mean value filtering method for digitizers. Background technique [0002] Bandwidth, sampling rate and memory depth are the three core indicators commonly used to evaluate the performance of digitizers. In addition, there is another very important indicator: vertical resolution. Higher vertical resolution means finer waveform display and more accurate signal measurement. The vertical resolution of a digitizer depends on the number of quantization bits in the analog-to-digital converter (ADC), as well as the noise and distortion levels of the digitizer itself. Although the number of bits in the ADC is commonly used to describe the vertical resolution of a digitizer, the exact parameter is the digitizer's overall system effective number of bits (ENOB). ENOB is closely related to the signal-to-noise-to-distortion ratio (SNR), and the ...

AI Technical Summary

Problems solved by technology

[0007] (1) Can only be used for periodic signals;

[0008] (2) Averaging can only reduce random, uncorrelated Gaussian white noise, and has little effect on correlated noise and interference;

[0009] (3) The averaging algorithm will significantly reduce the oscilloscope waveform update rate

For example, if you want to improve the resolution from 8bit to 12bit, you need to average every 256 acquisitions, and the waveform update rate is reduced by more than 256 times;

[0010] (4) Averaging may cause waveform distortion

[0012] (1) High sampling rate and high vertical resolution cannot be achieved at the same time;

[0013] (2) Oscilloscope bandwidth and vertical resolution cannot be combined;

[0014] (3) The high-resolution capture average algorithm is effective in suppressing Gaussian white noise, but has no effect on the noise caused by the INL (integral nonlinearity) of the oscilloscope ADC

For the collected samples containing gross errors, if they are directly averaged without processing, the result will deviate greatly from the real signal

Images