Simulation of improved algorithm of identification nd complications that arise during operation of complex technological systems

Abstract

Having analyzed foreign and domestic literature, it concluded that the most appropriate solution to the
problem of identification of problematic situations in complex technological systems is the use of methods and
algorithms of clustering. But every such algorithm has several features that affect its performance in this situation.
Therefore, to identify the current state of technological objects in information uncertainty it is recommended to use
neural classifier based on hybrid neuronet consisting of a network Kohonen and feed-forward neuronet. This
approach allows using control methods, in the case known statistical sample, containing data about the values of
information parameters and corresponding technical state of the object, identify and summarize the causal
relationship between the relation of the values of the information parameters of controlled object and its technical
condition.
The simulation of the proposed method was performed in Matlab medium, using SOM Toolbox, on the example
of technical state of rock cutting tool during drilling.
The control algorithm based on Kohonen neuronet can be improved and used to identify the different types of
technological object, as well as forecasting emergencies and complications that may arise during the operation of
complex technological systems.