The construction of real-time early warning model for drilling risk based on karst strata

Abstract:

Faced with various uncertainties existing in the drilling process of karst areas， this paper takes an engineering project in the karst of Western Guangdong as an example， relying on the long short-term memory （LSTM） model data prediction， and builds a drilling risk assessment framework in line with the local geological conditions， which can dynamically track and precisely predict drilling risks. It can be seen from the experimental data that the accuracy of hidden danger recognition of the model reaches more than 90%， which has important application value in ensuring the safety of drilling in karst areas， promoting the technological innovation in the field of drilling risk management， and providing a technical path for engineering safety protection under complex geological conditions.