Filipkovskaya L. Structural - analytical models, algorithms and software for recognition of industrial situations to varied tags

The object is productions of a different kind requiring classification data processing (CDP) in the control tasks; the aim is- a rising of a productions control efficiency by means of optimum models and tool of CDP; the methods are structural - analytical (SA) method of a pattern recognition, statistical analysis, theory of the tests, graph theory; theoretical results are new SA models of a pattern recognition, which one allow for the structural and analytical information on regularities of data domain, many-classes situations and varied tags of classification objects in productions control, and the statements, which one will be used for an estimation of quality of the classification rules; practical results are tools of information technique for automation of forming of SA models of classification of industrial situations, that allow to lower expenditures on control; novelty is optimum SA models of a pattern recognition on the basis of criterion of minimization of empirical risk and a procedure of selection of the control solution; a degree of implantation is universal; it is recommended to use in systems of handling of knowledge for technical training of production.