V.M. Belenkiy - Ph.D. (Eng.), MGU TU
U.V. Prus - Dr.Sc. (Phys.-Math.), Academy of State Fire Service

In this article we present an approach to calculation of the residual occupational hazard and of the social economic damages depend on influence of harmful and dangerous labor conditions. We expound the methods of events complex optimal planning which is used in computer-aided occupation safety control system at industrial enterprises. Based on cybernetics principles for building control systems, we present an approach to creating an occupation safety control system, containing data and knowledge base, a model that establishes a link between labor conditions and occupation hazard indicators, and a sub-system for planning optimal solutions. A process map for technological information analysis is provided. A model for risk identification is described. The model uses one of the methods of factor analysis (the method of principal factors) in combination with binary multivariate regression model. The adequacy of the model is evaluated based on statistical criteria, which allows risk indicator forecasting for a user-defined confidence interval. The system intends for optimal planning of prophylactic measures at labor safety. We discuss a mathematical model for correlation between injury, illness and a set of working factors. Offered functional structure automated managerial system by guard of labor in enterprise. The system is intended for optimum planning preventive action on guard of labor. The model describes the input and output information and control rules. The methodology for identifying indicators of occupational hazards based on labor conditions is discussed. The methodology includes developing probability models, eliminating insignificant factors, and calculating risks based on the values of user-defined arguments. The article provides a comparative analysis of statistical models that identify occupational hazards, such as factor analysis, multiple linear regression and neural networks, and justifies the choice in favor of the latter.