Abstract: Landslides,as a sudden and highly destructive geological hazard,pose severe threats to the safety of human production and livelihoods.The limited capability of single sensors to recognize multi-factor coupling effects hinders the comprehensiveness and accuracy of landslide early warning systems.To address this limitation,this paper proposes a multi-sensor fusion early warning model based on the BP neural network.Leveraging the nonlinear feature extraction capabilities of the BP neural network,the data from inclinometers,GNSS displacement sensors,and rainfall sensors are trained and predicted individually.The normalized predictions from these three sensors are then integrated using a weighted scoring method to achieve the final landslide risk assessment,forming an efficient and accurate monitoring system.The proposed early warning system has been successfully applied to a specific slope near the a certain oil pipeline,demonstrating promising results and significant potential for broader applications.

Key words: landslide monitoring, multi-sensor data fusion, BP neural network, risk scoring, geological hazard detection