USING THE NORMALIZED RANGE METHOD TO ASSESS THE IDENTITY OF TEST CYCLES

Abstract

The measurement accuracy of a microprocessor-based physical measurement sensor is
measured to the extent determined by its conversion characteristic, which is constructed from data
obtained from various calibration tests. Characteristics of the sensor converter, on which the
measurement accuracy depends, within a certain degree of approximation of the characteristics of
the sensor converter. Sensor calibration tests are carried out in accordance with the test procedure.
In the process of fulfilling obligations, errors arise due to exceptions of individual loops
made on each other. Therefore, small deviations from the conduction scheme can lead to a decrease
in the quality of the conversion characteristics and a decrease in the metrological characteristics
of the sensor. It is important that the results of several test cycles under constant environmental
conditions are independent of each other. The article presents a method for determiningthe quality of the results of calibration tests of a microprocessor pressure sensor. The method
allows you to evaluate previous test cycles for violation of the conditions of their conduct. An artificial
time series constructed using test data is analyzed. For construction, a specialized procedure
was implemented for connecting individual cycles into a single structure, similar to a time series.
A separate time series was constructed for each constant temperature value. Since the resulting
time series is a linear function, its Hurst exponent should be close to one. In this case, the series is
trend-resistant, and experimental cycles check independence and calculate a single linear trend
with minor deviations from it. If during the test any conditions for their conduct were violated, for
example, the conditions for transition from one temperature regime to a procedure, then based on
the results of the current test regime, the temperature conditions of the cycle regime will be observed.
To determine such scales, a procedure is proposed for comparing the Hurst exponent of a
time series, which presents data from an unreliable test cycle, with a range of acceptable results.
If the Hurst exponent meets the established limits, the test results can be used to construct a highquality
calibration characteristic. Otherwise, the results of the analyzed cycle describe the test
cycle conditions and recommend repeated test cycles.