Study On Evaluation Method For Measurement Uncertainty Of Virtual Instrument Algorithm

Due to the flexibility of virtual instrument it has been widely used in the industry, its accuracy directly related with the performance of the whole system.Uncertainty of the sensor, ADC has been relatively mature evaluation system, algorithm of uncertainty there is no quantitative data give good evaluation basis. Fast Fourier algorithm is widely used in virtual instrument, try to evaluate the uncertainty evaluation method of fast Fourier algorithm, this paper mainly completed the following work:(1) The measurement uncertainty degree theory are studied and summarized the characteristics and applicable conditions of the different measurement uncertainty, on the basis of that give the method of round-off uncertainty and truncation uncertainty evaluation.(2) Analysis fast Fourier algorithm, and its expression is decomposed into a sparse matrix, the additions and multiplications of each level are summarized. As computer data processing is often based on numerical rounding rules, based on the uncertainty are summarized in each transfer level, a common round-off uncertainty method can be established.(3) In the sampling of the signal, spectrum leak due to improper selection of sampling period and it will bring uncertainty. If sampling is synchronous, i.e., sampling period number to an integer, then the leak point of amplitude spectrum in the integer harmonic is zero, not cause analysis error. When sampling is synchronous, i.e., when the sampling period for the decimal number need to evaluate the uncertainty caused by leaking spectrum.The input signal hypothesis for the sinusoidal signal, according to the sampling period in the case of integer and non-integer the uncertainty of frequency spectrum leakage can be estimated.(4) Analysis the effect truncation of the periodic signal and the reason of the barrier effect to determine the frequency amplitude error, maximum error. Based on this error distribution system are estimated. Take triangle window function as an example, to calculate the maximum error of trigonometric function barrier effect and maximum error interval. Finally according to GUM, the truncation uncertainty can be calculated.(5) In order to verify the validity of the model, take the infinite periodic signal processing as an example. Analysising the sampling, truncation, fast Fourier transform precessing and calculating round-off uncertainty, spectrum leakage uncertainty, truncation uncertainty. And the three above synthesized obtain whether the virtual instrument algorithm uncertainty can be ignored.