А. S. Shaburov - Ph.D. (Eng.), Associate Professor, Department of Automation and Telemechanics, Perm National Research Polytechnic University E-mail: shans@at.pstu.ru R. B. Rashevskiy - Student, Perm National Research Polytechnic University E-mail: roman@rashevskiy.com

The topical issue of information security is necessary for security of information and control systems of critical objects. Apparently the most vulnerable objects are energy and water supply systems, transport management systems, banking systems and nuclear power complex elements. Arrangements for the information security which are currently in use are inefficient to confront modern threats. As the base for construction of information security systems the application of neural network has the most probable potential. At the present day it does not seem possible to achieve the effective operation of these systems in the automated workplace, for this reason it is offered to use the «client/server» architecture. This usage permits the sift of resource-intensive computations and analytical functional from the automated workplace to the central computation module. The central computation module is a hardware device which uses the neural network and accepts information from client agents, automated analysis and decision making, as well as it sends reference orders to client agents. Client agent is a software tool used as a filter for the operating system kernel. It captures references to the operating system kernel with regard to all possible acts of any subsystem. The item to analyze the validity of information exchange is the usage of fuzzy logic which considers the great amount of various factors. The advantage of this decision is the opportunity to implement hardware and software suite without the critical object information infrastructure modifying. The implementation of such hardware and software suite will be instrumental in information security threats and help to prevent highly dangerous situations.