Machine Training of the System of Functional Diagnostics of the Shaft Lifting Machine

The aim of the work is to increase the accuracy of functional diagnostics of a mine hoist by using the method of information-extreme machine teaching with a hierarchical data structure. The tasks set forth in the work were to develop a categorical model; to carry out synthesis based on its hierarchical machine teaching algorithm for a functional diagnosis system; and to optimize the system of acceptance tolerance. Functional diagnostics necessitates the analysis of a large number of diagnostic features and recognition classes that characterize not only possible malfunctions, but also intermediate technical conditions of nodes and assemblies of a complex machine. The proposed algorithm is developed in the framework of the so-called information-extreme intellectual data analysis technology based on maximizing the information ability of the system in the process of machine teaching. The main idea of the proposed method is to adapt the input mathematical description of the functional diagnostics system to the maximum reliability of diagnostic solutions in the process of machine teaching. The implementation of the proposed method of the information-extremal machine teaching is carried out by the example of functional diagnostics of a multi-rope mine hoist. The most significant result is the increase in the reliability of diagnostic solutions when using the hierarchical machine teaching algorithm of the functional diagnostics system as compared with the linear classifier. In addition, the crucial rules based on the optimal geometrical parameters of hyperspherical containers of recognition classes make it possible to take highly reliable diagnostic decisions in real time