Search results

The paper aims to develop a method for improving the accuracy of smart sensors (deemed as digital measuring instruments) by organizing combined measurements and processing their results by the parametric adjustment method at heterogeneous dispersion of the random error of the applied regression model.

When carrying out combined measurements, the problem of joint processing of measurement results of functionally related quantities must be solved. The function type can be known in advance or obtained experimentally. The number of combined measurements exceeds the number of unknown measured quantities. The redundant measurements can improve the accuracy of estimates of measured values but lead to inconsistency of the measurement results. The problem of inconsistency is solved by the parametric adjustment method, which is rather widely used mainly in the field of geodetic measurements, wherein the parametric equations are linear and the measured quantities are additive.

The proposed method allows to reduce the uncertainty of type B of a measurement result, caused by the maximum permissible error of a digital measuring instrument, by 1.2–4 times in comparison with the direct estimation method.

A compact description of the parametric adjustment method in matrix form is given. Recommendations are given on shaping a sensitivity matrix of functions for the proposed method. The geometric interpretation of the proposed method is considered. The results of the proposed method experimental testing are given when evaluating resistance values.