The object of the research thesis is information technology of forecasting landslides processes. The aim of the research is to increase the effectiveness of management decisions in planning anti-measures by developing models, methods and tools for monitoring and forecasting landslides of Southern Coast of Crimea. Research Methods. In the dissertation thesis for the research of the task used elements of a systematic approach, a method of retrospective studies, mathematical modeling, regression, correlation analysis and probabilistic graphic analytical methods of calculations in the form of Bayesian trust networks. The practical significance of the results: 1. There were developed methods, models and tools for monitoring and forecasting landslides processes of the Southern Coast of the Crimea, allows choosing informative indicators of the entire range of collected data that affect the activation of landslides processes of the Southern Coast of the Crimea, and by the proposed methods used to evaluate the landslide processes, thereby helped reducing forecast error. 2. Received opportunity to assess the prognosis of landslides SCC based on multifactor lagged autoregressive models and Bayesian trust networks, thus improving the efficiency of solutions produced by the introduction of anti-measures. 3. Obtained an opportunity to solve the problems of prediction and expert evaluation of complex objects and stochastic processes on the basis of developed and implemented intelligent system for monitoring and forecasting landslides SCC. 4. Based on the developed intelligent system for monitoring and forecasting landslides of SCC, received an opportunity to perform calculations of the effective level of costs of measures to prevent the catastrophic consequences of landslides SCC and the restoration of roads, buildings, structures, etc. 5. Theoretical and practical results used in the educational process at the European University in teaching courses: "Modeling systems", "Mathematical foundations of data analysis", "Automatic data processing", "System analysis and design of complex systems" and so on. 6. Theoretical and practical results of work implemented to the usage and adapted to the program library in Yalta MIS for information and computer center, thus improving the efficiency of designed solutions. Scientific novelty of the results: 1. For the first time was used a comprehensive correlation analysis of geological and climatic factors affecting landslide processes on Southern Coast of the Crimea, established their significance and made ranking. It became possible to conclude that little data impact of previous years, except for the values of rainfall, on the adequacy of modeling of landslides on the South Coast and the quality of their prediction. 2. For the first time was developed mathematical model for short-term forecasting and evaluation of landslide processes on SCC based on lagged multiple regression models with autoregressive component, which is simple and allows to increase the accuracy of short-term prognosis. 3. Improved information simulation technology and operational forecasting of landslides processes on the Southern Coast of the Crimea through the use of long-term observational data and expert estimates, which differs with full data analysis and for the first time takes into account the perturbation of independent factors via Bayesian trust networks. 4. Improved the technology of decision-making in determining and implementing anti measures proposed on the basis of models, methods and surgical short-term forecasting of landslides on SCC and determine the time horizon for which the catastrophic shifts may occur. This enabled us to evaluate the distribution of resources necessary to address damage from landslides and increase the effectiveness of anti-shits events in general. 5. Received further development the Box-Jenkins method of estimation and forecasting models to include criteria for evaluating the quality of the forecast absolute error of belonging to an interval not exceeding the half of standard deviation or exceeding him, relative error does not exceeding 10% or 50% exceeding, which allowed objectivize the selection process of models and increase the reliability of outcome prediction. Scientific and technical novelty of the results of research and publications and confirmed by a certificate of Ukraine for copyright on computed program.
