Klymov I. Methods and models for emergency risks monitoring and optimization

The object of study - the processes of emergency appearance and disaster risk reduction. The aim - to develop methods and models for monitoring and optimizing risks which minimize losses due to disaster appearance both of nature and technical kind by using process of knowledge mining and knowledge exchange while monitoring. Research methods - systematic and statistical analysis - for discovering nature of input data for processing and recovery processes, methods of expert evaluations, mathematical analysis, heuristic and mathematical programming - for solving the optimization problem in a distributed system of situational centers dedicated to risk management, methods of multiobjective optimization and decision theory - to build a model of monitoring centers interaction and synthesis models of emergency risk reduction. Equipment - personal computer. Theoretical and practical research results - an actual scientific task of minimizing losses due to appearance and expanding of natural disasters and hazards is solved by improving decision support systems for risk control systems and reusing mined knowledge obtained by knowledge base exchanges in monitoring systems for finding rational solutions within current technical limits; proposed methods and models are implemented as independent software module; Scientific novelty - developed the method of risk optimization, which is based on the use of a distributed knowledge base and stochastic estimates based on previous and current observations that can improve decisions on the amount of potential losses due to adaptive reuse of previous decisions, first proposed model of interaction for emergency monitoring centers based on geoinformational context and usage of self-organizing networks, which provides knowledge exchange capabilities; further development of situation model representation as composition of microsituations by considering connections between them, based on the use of a distributed knowledge base that enables increased performance optimization methods risk by allowing usage of distributed computing for problem solving; improved processing method and restore missing input measurement data by preliminary statistical analysis, which unlike analogues considers precedent geoinformational context which allows to increase the accuracy of reconstruction of the input measurement data. The results of the thesis as a software tool that implements the method of processing and recovery of missing input measurement data and the method of risk optimization in distributed network monitoring centers of natural origin, used in the joint venture "Computing machinery and automation systems" (Russia, Moscow) and JSC "Navidev" (Ukraine, Kharkov). Scientific and practical results of the thesis can be used: in the monitoring centers Emergencies Ministry of Emergencies of Ukraine for effective decision-making under multicriterial situations with incomplete data and, in the educational process for training specialists in the fields of designing systems for the prevention of disasters.