The measurement uncertainty of the measuring projector

The measurement uncertainty is refers to "the characterization of reasonable given the value of the measured dispersion, parameter associated with the measurement results ".

The definition of "reasonable ", which should take into account the correction by the effects of various factors on the measurement, especially the measurement should be in statistical control state, namely in stochastic control process. That is to say, the measurements are in repeatability conditions (see JJG1001 1998 " general terms and definitions for measurement of " fifth 6, this * * * shall refer to the standard terms ) and reproducibility conditions (see 5. 7 ) under the same measured, at this time to do multiple measurement, dispersion of the measurement result can be 5 . 8 of the Bessel formula, Sr and reproducibility standard ( partial ) differential sR and repeatability standard ( partial ).

The definition of " connected ", refers to the uncertainty of measurement is a measurement result " in the parameters together ", the result of the measurement (see article 1 of 5 ) a complete representation should be included in the uncertainty of measurement.

The uncertainty of measurement from understanding the meaning, means that the credibility of the measurement results, the degree of effectiveness of the doubt or uncertainty degree, is a parameter of quantitative description of the quality of the measurement results. In fact, because of the lack of understanding of measurement is not perfect and the people, the measured value with dispersion, i.e. each measured results are not the same value, many values but with a certain probability distributed in an area. Although the system error exist is a constant value, but because we cannot fully cognitive or master, only that it is a probability distribution in a region, and this probability distribution itself with dispersion. The uncertainty of measurement is that the measured values of dispersion parameter, it does not explain the measured result is close to the true value.

In order to characterize the dispersion, the uncertainty of measurement standard ( partial ) differential representation. In actual use, often hope confidence interval, know the measurement results therefore, in the definition of 1 states: the measurement uncertainty is also available standard ( partial ) differential ratio or the confidence level of the interval width representation. In order to distinguish these two different representation methods, they are for the standard uncertainty and the expanded uncertainty are known.

In practice, the measurement uncertainty may be derived from the following 10 aspects:

( 1 ) for the definition of the measured incomplete or imperfect;

( 2 ) method to achieve defined by measurement is not ideal;

( 3 ) the sampling representation is not enough, the definition of a sample is not representative of the measured by measurement;

( 4 ) under the influence of environment on the understanding of the measurement process is not complete, or measurement and control of the environmental conditions of the imperfect;

( 5) to simulated instrument readings are offset;

( 6 ) measuring instrument resolution or discrimination is not enough;

( 7 ) to measure value and reference materials ( CRMs ) values are not allowed;

( 8 ) refers to the data calculated constants and other parameters are not allowed;

( 9 ) the measurement methods and procedures of the approximation and assumption;

( 10 ) on the face of it completely under the same conditions, by measuring changes in repeated observations.

Thus, the measurement uncertainty from randomness and fuzziness, the former is attributed to the conditions are not sufficient, the latter is attributed to the thing itself is not clear. This makes the measurement uncertainty is generally composed of many components, some of which can be measured by weight results column ( observations ) statistical distribution was estimated, and experimental results of standard ( partial ) differential (see 5. 8 ) characterization; and some other component can use other methods ( according to the assumed probability distribution experience or other information ) was estimated, and also with standard ( partial ) differential characterization. All of these components, should be understood to have contributed to the dispersion. If you need to indicate that a component is caused by some reasons, can cause the random effect of uncertain cause and effect of the uncertainty of the system, instead of " random uncertainty " and " uncertainty " of the two already outdated or obsolete terminology. For example: the component of uncertainty caused by the correction value and the measurement standard, uncertainty can be called the system effect caused by.

Uncertainty when the variance obtained, take the positive square root. When the dispersion size with that level of confidence interval width when expressed as interval, the half width is obviously meaningless. When the uncertainty is divided by the measured results, called the relative uncertainty, it is a dimensionless quantity, usually expressed as a negative power percent or 10.

The uncertainty of measurement in the process of development, people from the traditional understanding of it is " characterization ( or description ) is the true value of the range of an estimate ( or parameters) "; also has a period of understanding " given by the measurement results is measurement estimation error of measurement may ". These have been used to define, conceptually is a development and evolution process, which relates to the true value and the measurement error of measurement is the two ideal or theoretical concept ( is actually unknown difficult operation ), and can be the specific operation is the change in the definition of the measurement results are that is, the value of the measured dispersion. As early as in the early seventy's, the international measurement have been more and more recognized by the use of " uncertainty " instead of " error " is more scientific, since then, the uncertainty of the term gradually in the measurement field is widely used. In 1978 the International Bureau of weights and the experimental uncertainty recommendation INC-1. In 1993 formulated the " guide to the expression of uncertainty in measurement " is BIPM, OIML, ISO, IEC, IUPAC, IUPAP, IFCC

Seven international organization approved, published by ISO, is an important authority documents of international organizations. China has also promulgated in 1999 measured compatible with the evaluation and expression of uncertainty in measurement technology specification. Thus, the evaluation of uncertainty of measurement has become one of the indispensable testing and calibration laboratory work. Because of the uncertainty of measurement theory is relatively new, there is a certain degree of difficulty in understanding. This paper discussed some characteristics of uncertainty.

One, the measurement results is an area

The purpose of measurement is to determine the quantity being measured value. The measurement results of quality is the measurement results of credible degree is the most important basis. The uncertainty of measurement is the quantitative characterization of the quality of measurement results, the measurement results of usability greatly depends on the size of the uncertainty. Therefore, the measurement results expression must contain both assigned to the measured value and the measurement associated with the uncertainty, it is complete and meaningful.

Characterization of assigned to the measured value reasonably parameter associated dispersion, with the result of a measurement, called measurement uncertainty. Dictionary of uncertainty ( uncertainty ) is defined as " change, uncertainty, unknown, uncertain ". Therefore, a broad sense, the uncertainty of measurement means dependability, measurement results of doubt or uncertainty degree of effectiveness of the degree. In fact, because of the lack of measurement is not perfect and people's knowledge, the measured value with dispersion, i.e. each measured results are not the same value, more value but with a certain probability distributed in an area. Although the system error exist is a relatively definite value, but because we cannot fully cognitive or grasp it, but that it is a probability distribution in a region, and this probability distribution itself with dispersion. The uncertainty of measurement is a description of the measured values of dispersion parameter, the uncertainty of measurement result reflects people on the measured value accurately understand deficiencies. Even after correction of the system errors of the measurement results, an estimate is still measured value

Value, this is because, random effect not only exists in the measurement will have uncertainty, and incomplete, correction system effect also exists uncertainty.

The original flow transferring system of standard device level allows the error must be less than the next level of standard 1/2 ~

1/3, uncertain development theory that we recognized the uncertainty of measurement result according to the evaluation method of uncertainty analysis, when the measured instrument repeatability is very good and the measurement process is better controlled, level two standard uncertainty of the difference may not much difference between, thus greatly reducing the accuracy transfer process the loss, the value transfer system is more reasonable.

Two, uncertainty and error

Probability theory, linear algebra and integral transform is the mathematical basis of the theory of error, after decades of development, the theory of error has been a system. Experimental standard deviation is the basic method to analyze the error, is not the basis for determining the degree theory. Therefore said the uncertainty theory is based on the theory of error from the essence, the basic analysis and calculation method is common. But the existence of relatively large differences in concept.

The uncertainty of measurement show that gives the value of the measured dispersion, is an interval obtained through analysis and evaluation of the measurement process. The measurement error is the difference of measurement results show that the deviation from the true value. The revised results may be very close to the true value (i.e., the error is very small ), but due to the lack of knowledge, people give it a value falls in a larger range (i.e., the measurement uncertainty is larger ). The uncertainty of measurement and the measurement error in concept, there are many differences.

Three, the uncertainty evaluation and type B evaluation of

With the statistical observations of row analysis assessed the standard uncertainty is called a standard uncertainty, different from the statistical analysis of observation column to evaluation of standard uncertainty is called type B standard uncertainty. The uncertainty is divided into " A" and " B", only for the convenience of discussion, does not mean that there is essentially difference between two kinds of evaluation, type a uncertainty is

The probability density function of frequency distribution derived by a group of observations obtained: type B uncertainty is the trust degree of occurrence of an event based on. They are based on the probability distribution, and use the variance or standard deviation characterization. Two types of uncertainty does not exist then a class of more reliable problem. In general, a ratio of B is more objective, and has strict statistically. Measurement of independence, is in a state of statistical control and measurement of the number of decisions type a uncertainty reliability.

" A ", " B " two kinds of uncertainty and " error " and " system error " classification no simple relationship exists between. " Random " and " system " that the two different nature of the error, " A" and " B " class representation of uncertain two different evaluation method of degree. Synthesis of random error and system error is not determined by the following principle, caused to the experiment results processing and confusion. And the type a uncertainty and type B uncertainty in synthesis using standard uncertainty, which is of uncertain progress degree theory.