Resumen
The subject of research is accurate temperature measurements by thermocouples whose legs undergone degradation during long term operation at high temperatures. The degradation rate is approximately proportional to the temperature of constant operation of the sections of thermocouple legs and time of operation. Degradation causes two types of errors ? due to drift of the conversion characteristic (its gradual change during operation) and due to the acquired thermoelectric inhomogeneity of thermocouple legs (manifests itself as a change in thermocouple conversion characteristic when changing the temperature field even at constant temperatures of the measuring and reference junctions). The aim of the article is to study the method of increasing the accuracy of temperature measurement by thermocouples that have significant thermoelectric inhomogeneity acquired during operation. The aim is achieved by stabilizing the temperature field along the thermocouple legs, so error due to acquired thermoelectric inhomogeneity cannot manifest itself. The tasks of the paper are to estimate the error of temperature measurement using the thermocouple, develop the method for increasing accuracy of temperature measurements by inhomogeneous thermocouples, means of stabilization of the temperature field, methods of control of the temperature field, and experimental study the proposed control methods. There are well-known methods used such as the design of equipment with given parameters and limitations, calculations by solving a system of linear equations, the formation of the necessary functional dependence using a neural network, experimental studies. The obtained results are as follows: stabilization of the temperature field makes it possible to reduce the influence of acquired thermoelectric inhomogeneity of thermocouples, which, according to the literature, can reach 11°? to 1,3°?. It is shown that the developed multi-zone tubular furnace for stabilizing the temperature field due to a close thermal connection between the zones, requires a multi-channel controller that is not prone to self-excitation. There are two methods of control of the temperature field are proposed. One based on the solution of a system of linear equations. The second method is based on the neural network, in which the neural network is trained directly on a multizone furnace. Conclusions. As shown by experimental studies, the proposed methods do not allow the error of the acquired thermoelectric inhomogeneity of thermocouples to manifest itself, which provides an opportunity to increase the accuracy and metrological reliability of temperature measurements by existing types of thermocouples.