Metrology of nerve-similar sensing systems in anthropomorphic robots and living organisms

Abstract

The features of metrological support of nerve-similar sensor measuring systems with the use of neural networks are considered. It is established that the metrological verification of such measuring systems, by analogy with living organisms, is possible by comparing their responses, invariant to the external unstable environment. It is shown that from the standpoint of classical metrology, instrumental, methodical, static, dynamic, systematic and random errors of nerve-like sensory measurement systems with neural network results are reduced to two types – methodical and systematic errors of neural network training. It is proved that without using of reference calibration tools for embedded sensors, it is possible to experimentally evaluate systematic and random errors in the presence of several similar measurement systems by multiple pairwise comparisons of their responses, and the obtained systematic error can be introduced in the form of a metrological correction in the results of responses.