Using the data of the national spatial data infrastructure in Ukraine for the study of risk assessment of critical assets is one of the most important applied problems. Relevant and comprehensive spatial data about climatic conditions, engineering networks, and accident statistics should be provided by mapping services and local management companies for capabilities of decision support solutions and prediction emergencies.

Power industry is a one from basic industry in Ukraine and a strategic sector in any country. Power industry security is a component of country security in general. Hazardous situations (accidents) on the objects of energy power systems usually arise from defects in the manufacture and operation of the equipment, personnel rules violation and other factors and lead to the forced termination of energy supply, causing a threat to the life of society.

Primary part of the power transmission networks are the overhead power transmissions lines, because of this there is a threat of adverse impact of climatic factors on the power transmission network components. Extreme climatic conditions lead to accidents on the power lines, so the problem of analysis of climate impacts on the power transmission network and prediction the consequences of these effects are direct component of power system security problem. Extreme accidents analysis shows that more than half of the failures on overhead power lines caused by the ice and wind overloads on the wires, cables and other structures.

Bayesian network is used to simulate accidents on power grid objects. Bayesian network is a graphical model that encodes probabilistic relationships among studied variables. The graphical model has several advantages for data analysis: coding dependencies between all variables and easy handling situations when data are missing; studying the possibility of using cause-effect relationships; avoiding the need to "fit" the data.

Model development for accident under the influence of climatic factors takes following stages: defining models variables and relations between them; Bayesian network structure construction, determining the possible values of variables and a priori probabilities; Bayesian network learning and refining its structure (variables and their probabilities); model testing using accidents data at power lines and meteorological observations; prediction the occurrence of accidents involving information about the accident using the constructed model. Bayesian networks prediction is based on a Bayesian classifier, which is statistically optimal classifier, which minimizes the risk of misclassification. Bayesian network model was tested using the technique of cross-validation.

Variables that are used for accidents simulation are the following: ice weight, event duration and ice growth period, type of topography, the constructions lifetime, the wind speed at the maximum ice load period, month of ice load occurrence, altitude and wind direction at the beginning and after reaching the maximum size of ice. Model development is performed on meteorological observations, accidents cases data and geospatial data of power grid from SDI organizations network.

Zoning maps for the area of interest were constructed with QGIS, marking on the maps locations of accidents that have occurred, and predictable accidents place.