IEC_60060_-_2_High_Voltage_Test_Techniques_-_2_-_Measuring_Systems-_English.pdf

IeC60-2944844891061241212360-2 Amend.1 IEC:1996-5-An example of a measurement with its uncertainty is:1 040 kV 20 kV(estimated confidence level not less than 95%)All measurements are to some degree imperfect.A Measuring System will be influenced byvarious quantities(e.g.temperature,proximity of earthed or energised structures,interferenceetc.).It will generally be found that when a measurement is repeated several times there will bea spread in the results(when the requirements of this part of IEC 60 are met this spread willbe small).When a measurement is repeated a large number of times,most of the resultswill be close to a central value and this central value will tend to a constant value as thenumber of measurements is increased.Many high-voltage tests permit only a single measurement.Others require several measure-ments,e.g.10 as in clause 6 of this part of IEC 60.A single measurement may give any valuein the likely distribution.The possible differences between a single value(or the mean of asmall number of measurements)and the mean of the distribution of all possible values gives arandom contribution of uncertainty.This annex provides procedures to deal with any number of repetitions of a measurement.In most measurements,the overall uncertainty will result from a combination of severalcontributions which are classified into two categories according to the method used to evaluatetheir numerical values 1*.H.2.1 Systematic contributions(Type B 1)Systematic contributions are those which are not evaluated statistically but are estimated byother means.Some examples are:-the uncertainty of the calibration of the Measuring System(or its components),stated inthe calibration certificate;-the drift in the value of the scale factor of the Measuring System(e.g.ageing);-the use of a Measuring System under constant conditions which are different from thoseof the calibration(e.g.different temperature);-the resolution of each instrument.Once a Measuring System(or a component)has been calibrated and is then used in a test,theuncertainty of the calibration is treated as one of the systematic contributions in the estimationof the uncertainty of the test result.H.2.1.1 Systematic contributions(rectangular)It is assumed that these systematic contributions have a rectangular distribution,that is,anymeasured value between the estimated limits(a,where a is the semi-range value)is assumedto be equally probable.The figures in square brackets refer to H.7 Reference documents.Copyright nternational Electrotechnical CommissionSIEMENSrovided by IHS under license with IECTue Mar 30 16:49:09 MST 2004No reproduction or networking permitted without license from IHSIEC60-29448448910612414TT660-2 Amend.1 IEC:1996-7The.standard deviation of a rectangular distribution is:503a(H.1)When a number(n)of uncorrelated systematic contributions(rectangular)are combined 2,then:a)the standard deviation is:孟跪酯酯s=333(H.2)3where a to an are the semi-range values of the individual contributions;b)provided there is a sufficient number of significant contributions,the distribution isapproximately Gaussian.H.2.1.2 Systematic contributions(Gaussian)It is assumed that these systematic contributions have a Gaussian distribution.When a numberof uncorrelated systematic contributions(Gaussian)are combined the square root of the sumof the squares of the appropriate standard deviations is evaluated(ssg).H.2.2 Combination of systematic contributionsThe standard deviation for all systematic contributions is:56=V5品+喝(H.3)H.2.3 Random contributions(Type A 1)Random contributions are those which are derived statistically from repetitive measurementand,being random,will usually be found by measurement to tend to a Gaussian distribution.Each random contribution is characterised by the experimental standard deviation(s,)of thesample of measured values(see equation(H.11)in H.3.3.1).H.2.4 Correlation between uncertainty contributionsCorrelation between measured quantities should not be ignored if present and significant.Ifpossible,this correlation should be evaluated experimentally by varying the correlatedquantities.In many cases the measured quantities are sufficiently independent that the affectedquantities can be assumed to be uncorrelated.If it has been decided that the correlationbetween measured quantities cannot be ignored,and cannot be determined experimentally,then the procedures in ISO TAG 4,section 5.1,should be used 1.H.3 Overall uncertaintyH.3.1 Combination of uncertainty contributionsThe uncertainty for a Gaussian distribution is given by:U=ks(estimated confidence level of P%)(H.4)SeiseEaeteieEcommsion901aageTmIEC 60-2944844891061241687960-2 Amend.1IEC:1996-9-wherek is the normal distribution factor(coverage factor 1)and is given,for n-，in the lastline of table H.1;P is given in the first line of table H.1.Unless otherwise specified it is usual to evaluate uncertainty at a 95%confidence level and arounded value of k=2 is used 1.The procedure in this annex for the combination of uncertainty contributions requires that thesystematic and random uncertainty contributions of a Measuring System are evaluatedseparately.The derivation of the overall uncertainty U is based on the square root of the sumof the squares of the systematic and random uncertainty contributions:u=u+u(H.5)and Ug and U,are calculated at the same confidence level and are derived in accordance withH.3.2 andH.3.3.H.3.2 Estimation of systematic uncertaintyThe basic equation for systematic uncertainty U is:Us=kss=ks+s(H.6)(from equation(H.3)and equation(H.4).If a calibration uncertainty is given without any indication of confidence level it should betreated as having a rectangular distribution with a semi-range value equal to the uncertaintyand included as one of the terms in equation(H.2).When the uncertainty is given with a confidence level it should be assumed to have a Gaussiandistribution.Therefore,if the calibration uncertainty is given at a level of confidence of 95%asis now being generally recommended,the value is a 2 s value(i.e.k=2),and:U95Ssg=2(H.7)In particular,where there is a single calibration for the complete Measuring System,equation(H.6)becomes:(H.8)where a1 to an are the semi-range values.The general form of equation(H.6)is:(H.9)Cooright intematiceal Flectroechoical CmSIEMENSTue Mar 30 16:49:09 MST 2004reproduction or networking pemted without license rom HS