V. M. Belenkiy – Ph.D. (Eng.), Senior Lecturer, Information Technologies Departament, Moscow State University of Technologies and Control (MGUTU) named after K. G. Razumovskiy. E-mail: email@example.com
V. G. Spiridonov – Post-graduate Student, Moscow State University of Technologies and Control (MGUTU) named after K. G. Razumovskiy. E-mail: firstname.lastname@example.org
The paper describes: structure of the “Neural network predictor” program, its interaction with database created on the basis of database control system MS Access, implemented mechanism of entering new values of the neural network coefficients and the algorithm for obtaining these coefficients using the software package “Statistica 6”. “A neural network predictor” is used to predict the incidence of morbidity and injuries based on the enterprise labor conditions.
The calculation module of the program uses a universal algorithm of neural networks (NN), which automatically determines NN structure and type based on selected options (object of study – enterprise, type of illness, NN configuration), as well as corresponding NN coefficients. All the necessary information is transmitted to the program forms, which are created in DELPHI, from a database that is uniquely structured to store all of the information needed by the program (NN coefficients, disease type, labor condition factors, and the object of study). The database (DB) is designed based on optimum data storage principles. Thanks to the logical structure of the database, all of the necessary information is retrieved with the help of SQL-queries, while referential integrity is maintained in the event of change of data. All the necessary information in the database is selected from the initial statistical data and “Statistica 6” package. The package is used to analyze and process of the obtained information on morbidity and labor condition factors, using regression analysis and NN modules.
Use of this software as one of the modules of the system for optimal planning of safety measures allows to choose an effective set of measures aimed at reducing morbidity at an industrial enterprise.