Observational error

It is generally not possible to measure exactly right. Through a variety of causes the variable to be measured is not detected correctly. The deviation of a value obtained from measurements of the true value of the measurand is called the error of measurement (according to DIN 1319-1:1995 ) or measurement errors (old name). The key messages of this comprehensive standard for this keyword are explained below.

  • 4.1 Systematic measurement errors,
  • 4.2 Random measurement errors

Preliminary

Just the numerical values ​​of measured variables have uncertainties. One should always question in the indication of a result:

  • How far I can rely on the displayed ( calculated ) value as a correct statement of the quantity to be measured?
  • As far as I can rely on the established numerical value?

Definitions

True value and true value

In the basic for metrology DIN 1319, a distinction between these two values ​​:

  • = True value of the measurand as a target of the evaluations of measurements of the measurand; this is an " ideal value", which is not exactly known in most cases.
  • = True value of the measured variable as a " known value " for purposes of comparison, the deviation from the true value for the comparison purpose is considered to be negligible.

The correct value is the value that an error-free measuring device would spend a. By comparison with a standard determined (or fundamentally determined or as properly defined ) value Between and there is a true principle, but quantitatively insignificant difference.

Subsequently, the quantitative tangible value is used. together with

  • = Displayed ( issued ) value

Provides a comparison of with the absolute and the relative deviation of a measured value.

Absolute deviation

Or absolute error

This quantity has a magnitude and a sign of a unit, namely always the same as the measured variable.

Relative deviation

Or relative error

This size is always unitless; It may be positive or negative.

  • Note 1: A value does not give rise to a measurement.
  • Note 2: It is true: for all. In addition: 100% = 1

Example:

Sources of measurement errors

  • Meters deviations as a result of imperfection of the design, fabrication, adjustment ( eg materials, manufacturing tolerances)
  • By the measurement method -related effects due to the influence of the measuring device on the measured variable (eg retroactivity deviation [ circuit influence error ] through self-consumption of the meter)
  • Environmental impacts as a result of changes in impacts from the environment ( eg temperature, external electric or magnetic fields, position, vibration )
  • Instabilities of the value of the measurand or of the institution of the measured variable ( eg statistical processes, noise)
  • Observer influences due to different characteristics and abilities of the people (eg, attention, exercise, visual acuity, estimated assets, parallax)

Are outside the discussion here

  • Distorted by errors of the observer,
  • Distorted by selection of inappropriate measurement and evaluation,
  • Distortion caused by non-observance of known interference.

Types of measurement errors

Systematic measurement errors,

All deviations which are unidirectional and - though difficult - could be identified, are systematic deviations.

  • Systematic measurement errors have magnitude and sign.
  • Known systematic deviations are ruled out by rectification.
  • Unknown systematic errors can at best be summarized on the basis of appropriate experience in a component of the measurement uncertainty.

Random measurement errors

Uncontrollable, not uni-directional deviations are random deviations.

  • When reps - even under exactly the same conditions - the readings will differ; scatter them.
  • Random measurement errors vary in magnitude and sign.
  • Using a calculation error can be calculated from the total of the average values ​​and a component of the measurement uncertainty. The true value ( in the absence of systematic errors ) with a degree of confidence in one area.
  • The total measurement uncertainty is given by.
  • Through systematic errors a measurement result is always wrong.
  • Due to random errors a measurement result is always uncertain.

Error limit

The margin of error is conceptually clearly distinguished from error. It tells us how large the error may be in absolute value at most. In this case, there is an upper and a lower error limit, preferably the same size, described by the unsigned magnitude. The true value ( in the absence of random variation) in an area.

Occasionally it is possible to improve a measurement method and so to reduce the error limits; while there remains the question whether it is worth the increased (cost) effort.

In many areas of the MPE are the subject of regulations; then verification offices and industrial laboratories are specialized to deal with it.

Meters deviations

Since its preparation Each instrument contains instruments deviations. This can be determined by comparison with a much better gauge; so they are of a systematic nature and in principle correctable. The cost for this is however high. On dealing with the deviations there are two ways, one of which should be supplied by the manufacturer of the measuring instrument:

Error analysis
565525
de