A system approach to geodynamic zoning based on artificial neural networks

Abstract

Methodological aspects of applying artificial neural networks for geodynamic zoning of territories when selecting locations for environmentally hazardous facilities (as exemplified by nuclear fuel cycle facilities) are considered. To overcome the uncertainty caused by complexity of the analysis of information about the properties, processes, and structure of the geological environment, a system approach to analyzing the information is used. Geological environment is presented as a system of interacting anthropogenic facility and natural environment, between which connections are organized. When assessing safety of such system operation, it is important to monitor environmental condition indicators. According to modern regulatory requirements of both international and domestic organizations, one of the main, and at the same time difficult to determine indicators of the condition of the nuclear fuel cycle facilities sites, are modern movements of the earth's crust. In the paper, we presented a method for predicting modern movements of the earth's crust based on artificial neural networks. Based on the predicted kinematic characteristics of the earth's crust, it is possible to identify zones that are dangerous in the manifestation of geodynamic processes: tension, compression, elastic energy accumulation zones, and so on. Preliminary results obtained on the presented neural network architecture showed positive prospects of applying this methodology for solving geodynamic zoning problems.

Keywords

• artificial neural networks
• geodynamic zoning
• modern movements
• strains
• radioactive waste
• geological environment
• system approach