Introduction. In the branch structure of management of Russia an important place is occupied by geographically distributed objects (branches) of large industrial enterprises of the fuel and energy complex. The considered enterprises, in accordance with the criteria approved by the legislation, belong to the category of hazardous production facilities. An important task is to determine the cause of the danger. Methods of research. To detect hazards in any automated control systems used methods by which information is collected on the parametric values of the functioning of production facilities. To conduct research on the detection of hazards, a number of approaches are used: based to determine the parameters (invariants) of the models of controlled objects; to solve the problems of modeling (forecasting); to use analytical redundancy. There is a modelfree method of hazard detection in automated control systems, which is based on the representation of only the data of control signals and measurements of the parameters of the functioning of dynamic objects. It is based on the algebraic condition of solvability of the problem of identification of a mathematical model of dynamic object functioning. Problem statement. It’s required on the basis of the measurement results the input signals coming into the automated control system, to develop a parametric value for the critical zone of occurrence of hazards with the aim of displaying information on the display of the workstation operator. Problem solution. It is offered to represent models of object, in serviceable and faulty States, in the form of matrices that will allow solving problems of identification of the closed objects for any input signals, irrespective of availability of information on parameters of control system. The speed and accuracy of detection of the fact and time of danger (system failure) are determined by the sampling frequency of the signals and coincide with the time interval between the two consecutive measurements. Conclusion. The advantage of the proposed approach is its independence from the parameters of the controlled object model. The use of the proposed approach for detection makes it possible to transfer the security management system of the enterprise to a new qualitative level due to constant traceability of the process of functioning of production facilities, increasing the speed and reliability of detection of the fact and time of danger.