Ensemble of artificial neural networks to control the induction soldering of spacecraft's waveguide paths

This article discusses the development of an intelligent technology designed to solve the problem of controlling the process of induction soldering of waveguide paths of spacecraft. The solution to this problem is complicated by the peculiarities due to the use of non-contact temperature sensors. Intelligent methods have proven themselves well in solving control problems under conditions of uncertainty. The use of intelligent methods is well suited both for solving problems of identification and correction of errors in measuring instruments, and for direct control of the technological process of induction soldering of waveguide paths of spacecraft. The essence of the proposed technology lies in the use of ensemble of artificial neural networks to solve the problem of controlling the induction soldering process at its following stages: assessing the quality of temperature measurements in the heating zone obtained using pyrometric sensors; correction of measurements in case of detection of non-standard errors in measuring instruments; control of the induction heating process in the absence of reliable readings of measuring instruments. The use of the presented intelligent system will improve the quality of control of the technological process of induction soldering of waveguide paths of spacecraft by reducing costs associated with the features of the technological process and non-contact temperature measurement sensors.