Analysis of stable functioning of objects using machine learning

Abstract

Stable functioning of the technical objects is estimated using methods of the statistical process control. However this approach does not always provide the timely detection of violations. It is suggested using machine learning methods for the binary classification of object states (stable or unstable). A program has been developed for calculation in the Matlab environment which allows for analysis of impact of the learning method, classification quality criteria, method of validation set as well as methods of selection of significant indicators on the object’s stable functioning forecast precision. Stable operation of the water treatment management system, stable vibration of the hydraulic unit, machining operation process are taken as examples.