Оценка трещиноватости отрабатываемого горного массива на основе аэрофотосъемки и методов машинного зрения

Abstract

The task to assess rock mass jointing are currently usually solved in manual mode, which requires high qualification of the specialists and considerable time expenditures. Automation of such tasks is important in terms of reducing the time of image processing and obtaining additional information on the geomechanical state of the rock mass. The article discusses the possibilities of using computer vision and artificial intelligence technologies to assess jointing of the rock mass. For this purpose, aerial photography data obtained using unmanned aerial vehicles are used. The images are processed with the software developed by the authors, which performs tracing of the joints based on a neural network of a dedicated architecture. The results of processing aerial photography data are presented using the cases of coal strip mines in Kuzbass and open-pit mines of the Kola Peninsula. The use of neural network in processing of the aerial survey data of the rock masses has shown the promising potential of the method. After processing the data of tracing the jointing fields, it becomes possible to monitor the behavior of the rock mass by using the visualization tools for additional fields of characteristics, which allow to assess the nature of changes occurring under anthropogenic loads. The developed algorithms make it possible to significantly accelerate the processes of aerial survey data processing to assess the structural disturbance of the rock mass.