Susceptibility zoning of karst geological hazards using machine learning and cloud model

Accurate assessment on serious geological hazards using artificial intelligence is required in geological hazards assessment with aim to improve the efficiency and accuracy of karst geological hazard identification. In this paper, topography and geology characteristics are selected as susceptibility factors for geological disasters identification. The area of Guibei, Guangxi Province, was selected as the study area because of its typical karst landform. Intelligent classification of influential factors for karst geological hazard identification is realized by utilization of support vector machine, meanwhile, the cloud model theory was introduced and a comprehensive method of susceptibility zoning deployment of geological hazards in karst areas was proposed, further, a quantitative susceptibility assessment model for unsuitable geology in karst areas was established. After that, a comprehensive geological hazard susceptibility index (S) proposed in this paper is calculated with assistance of machine learning theory in the proposed model. The results show that the influential factors can be divided into five grades through statistical learning, the susceptibility index of geological hazards in the study area is between 0.713 and 5.798, the susceptibility of geological hazards in the mid-section take the largest share. Differences in the distribution of high, middle, and low susceptibility zones are significant. Different susceptibility zoning have different influence on engineering construction.