Procedure of construction of calibration dependence of measuring tools is examined by using artificial neuron networks, which allow to take the choice of character of dependence between value of measurement standard and corresponding indication of measuring tool either to linear or to nonlinear. A teaching order of artificial neuron networks by the method of back propagation error and by algorithm of Levenberg-Marquardt is described to applies to the construction of calibration dependence. Empiric comparison of results of calibration by the method of least squares and by neuron networks is conducted.
measuring vagueness, calibration of measuring mean, least-squares method, neuron network
“Ispolzovanie iskusstvennykh neironnykh setei pri postroenii kalibrovochnoi zavisimosti sredstva izmereniia”,
Information Processing Systems,