Kachur S. Control process automation of nuclear power installations based on Petri nets

Subject of research is control process of nuclear power installations. A research purpose is control process automation of nuclear power installations based on Petri nets' models which provide a necessary speed at emergency situations' localization and prevention to increase APP exploitation safety. Research methods are automatic and adaptive control; control of a nuclear reactor; simulation; Petri nets' theory; artificial intelligence; thermal physics and thermal hydraulics; applied thermodynamics; applied statistics. A new decision of an important scientifically-applied problem of nuclear power installations (NPI) control process automation by creating objects' models and control devices with a variable random structure on a set of stochastic elements when a structure set hasn't been specified yet, and by developing based on these automated control systems' methods which are characterized by high speed in emergency situations, have been got. Object models and NPI control devices described as probabilistic connections of stochastic systems based on Petri nets' expansion which makes it possible to reduce a process model size and to increase significantly process control speed of heat exchange in an active reactor zone; system controllability criterion described by Petri nets' models that reduces a dimension of an object controllability verification problem and increases speed of a nuclear reactor transition determination to an uncontrolled state; a neuron module model of variable quantity described by Petri nets as a superconductor with variable time activation which makes it possible to create a new element base for neural networks' hardware implementation and to increase modeling capability of software implemented neural networks; an approach to intellectual control of a reactor using a modular neural network where a possible modules' connection is taken into account that increases speed of making decisions at NPI emergency situations; a mathematical model of a vaporization process in a channel of a reactor active zone, which forecasts basic parameters of a heat exchange process and determines parameters of heat exchange crisis before the beginning of surface boiling in a technological channel which makes getting some additional time possible for making decisions by an APP operator in a case of an emergency situation or for operating of protection devices at an extraordinary situation have been offered for the first time. Petri nets' mathematical theory which executes maximum probability verification of marker transmission based on stochastic and functional Petri nets conjunction that allows to make a choice of the most probable structure based on a mathematical model; an identification method of reactor anomalies based on optical sensors' measuring which are located on a mixing chamber perimeter to determine iteratively spatial distributing maximums of steam quality that reduces processed information quantity and increases speed of diagnostics systems have been improved. The probability-theoretic methods of optimum statistical decisions for computer simulation tasks like filtering, identification and structure and parameter optimization based on Petri nets have been developed which increases solution speed of these tasks during NPI control process. Practical results of the research are description methods of a control object which were developed in the dissertation, object identification methods, intellectual control systems' creation, and an operator-oriented training system which increases NPI control safety. Results of the dissertation research have been applied at Zaporizhya APP and Yuzhnoukraynsk APP, at the research reactor IR-100 and they have been used in an educational process in Sevastopol national university of nuclear energy and industry. Scientific and practical results of the dissertation can be used for diagnostics and control systems of NPI APP creation and for different types of transport NPI.