Study On Evaluation Method Of Measurement Uncertainties Of Transducers

The measurement accuracy of transducers has a greatest impact on the measurement result of an instrument; therefore, it’s necessary to study the appropriate method to evaluate the accuracy of a transducer. One approach is to evaluate the measurement uncertainties of the transducer. Currently the most common methods are Type A and Type B methods based on GUM (Guide to the Expression of Uncertainty in measurement). According to GUM, if the sources of measurement uncertainties of a transducer are available, the combined standard uncertainty of the transducer can be obtained according to the propagation rules of GUM. However, with respect to this method, the measurement uncertainty result is not accurate enough because it views the synthesized distribution as normal distribution. As a result it is necessary to study a more accurate and evaluation method on condition that all the sources of uncertainty are available and they are independent of each other.In an attempt to solve the above problems, this paper has carried out the following studies:(1) The fundamental and the general principles of measurement uncertainty evaluation in GUM are analyzed; especially its application area is also discussed in detail.(2) Several methods about measurement uncertainty evaluation are also compared; including GUM method, evaluation of measurement uncertainty based on Type A of Bayesian theory, and evaluation of measurement uncertainty according to type B based on Gram-Charlier series. Their advantages and disadvantages are presented.(3) The sources of measurement uncertainties of transducers are analyzed; meanwhile the probability distribution of each source is studied; moreover the special conditions under which the combined probability distribution could be obtained are also discussed.(4) Once the sources of uncertainty of transducer are independent of each other, the evaluation of measurement uncertainties of a transducer could proceed by convolution approach based on FFT. The procedures are:1) determine the synthesis distribution of the of all sources of measurement uncertainty of a transducer with linear convolution; 2) convert the linear convolution to circular convolution, thereby obtaining the combined probability distribution; 3) determine the measurement uncertainty of a transducer by convolution approach based on FFT. This method has been successfully applied to evaluating the measurement uncertainty of a laser sensor FT50RLA-20-S-L4S, and its results have been compared with the GUM method.Once all the sources of uncertainty of transducer are independent, its static measurement uncertainty could be obtained by convolution approach based on FFT. Under condition that the same confidence level is specified, the convolution method proposed is more accurate than the method recommended by GUM.