374
P. Liu
Due to the limitations of the test equipment (wind tunnel, water tunnel,
etc.) methods, test environment, measuring instruments, measurement procedures, etc., there is always a certain difference between the measured value
and the real value of any physical quantity. Absolute error, relative error,
or significant figures are often used to illustrate the accuracy of an approximation. In order to assess the accuracy of the test data, it is necessary to
analyze the source of the error and its effects. Therefore, it is determined
which factors are the main aspects affecting the accuracy of the test, thereby
further improving the test plan in the test, reducing the difference between
the test observation value and the real value, and improving the accuracy of
the test.
1. Concept of error
(1) True value and mean value
The real value is a certain value that is objectively present in the measured
physical quantity, also called a theoretical value or a defined value. Usually,
the real value is unmeasurable. If the number of measurements is infinite in
the test, the probability of occurrence of positive and negative errors is equal
according to the law of distribution of errors. After careful elimination of
systematic errors, the measured values are averaged to obtain values that are
very close to the real value. But, in fact, the number of test measurements
is always limited. The average value obtained with a finite measurement can
only be approximated to the real value. The commonly used average values
are as follows: arithmetic mean, geometric mean, root mean square average,
and log average. It should be noted that the logarithmic mean of the variables is always less than the arithmetic mean. The purpose of taking various
averages is to find the one closest to the real value from a set of measured
values. In wind tunnel tests, the distribution of data is mostly normal, so the
arithmetic mean is usually used.
(2) Classification of errors
According to the nature of the error and the reasons for its occurrence, it is
generally divided into three categories:
(1) Systematic error: Systematic error refers to the error caused by factors
that are not detected or confirmed in the measurement and test, and
P. Liu
Due to the limitations of the test equipment (wind tunnel, water tunnel,
etc.) methods, test environment, measuring instruments, measurement procedures, etc., there is always a certain difference between the measured value
and the real value of any physical quantity. Absolute error, relative error,
or significant figures are often used to illustrate the accuracy of an approximation. In order to assess the accuracy of the test data, it is necessary to
analyze the source of the error and its effects. Therefore, it is determined
which factors are the main aspects affecting the accuracy of the test, thereby
further improving the test plan in the test, reducing the difference between
the test observation value and the real value, and improving the accuracy of
the test.
1. Concept of error
(1) True value and mean value
The real value is a certain value that is objectively present in the measured
physical quantity, also called a theoretical value or a defined value. Usually,
the real value is unmeasurable. If the number of measurements is infinite in
the test, the probability of occurrence of positive and negative errors is equal
according to the law of distribution of errors. After careful elimination of
systematic errors, the measured values are averaged to obtain values that are
very close to the real value. But, in fact, the number of test measurements
is always limited. The average value obtained with a finite measurement can
only be approximated to the real value. The commonly used average values
are as follows: arithmetic mean, geometric mean, root mean square average,
and log average. It should be noted that the logarithmic mean of the variables is always less than the arithmetic mean. The purpose of taking various
averages is to find the one closest to the real value from a set of measured
values. In wind tunnel tests, the distribution of data is mostly normal, so the
arithmetic mean is usually used.
(2) Classification of errors
According to the nature of the error and the reasons for its occurrence, it is
generally divided into three categories:
(1) Systematic error: Systematic error refers to the error caused by factors
that are not detected or confirmed in the measurement and test, and