Математическое моделирование процесса принятия решения о состоянии стохастических систем

Because of the difficulty of constructing rigorous mathematical models of technological, biomedical and economic facilities have been developed methods of forecasting based on statistical analysis. The complexity of the analyzed object is equivalent to its information capacity. The maximum capacity is achieved if all of the object’s state is equally likely. The relative uncertainty of information obtained crucial system complicates decision-making on the state of the object. To reliably predict the state of the object is measured several characteristics, the measurement range is broken into grades, but within each gradation is made by averaging of the signal. Next solve two problems: the detection problem (detection of deviation of operation from normal mode) and a recognition task (assessment of the degree of deviation from the norm). The number of gradations of the trait is closely linked to the capacity of the training sample (at least 40). In the description of the system from 8 to 30 signs and power training samples from 40 to 120, the method includes the formalization of the signs in the first stage, the selection using a correlation analysis of the most informative features in the second stage and a classification of the state of the object by the method of cluster analysis allowed to correctly diagnose the system status in emergency mode with an accuracy of between 89 to 98%. The proposed information approach allows the classification and prediction of technical, economic and biomedical systems of any complexity, which opens up the possibility of predicting the behavior of such systems and control the appearance of interference.