Intelligent analysis and early warning technology study on monitoring data of reservoir dam based on improved 3-sigma hybrid model
The determination of the safety threshold range for monitoring data of reservoir dams and the construction of an early warning system are key factors for dam safety monitoring.The main task of formulating early warning indicator thresholds is to evaluate and predict the ability to resist potential loads based on the dam's ability to withstand experienced loads,in order to determine the extreme and warning values of the monitoring effect under this load combination.This technology is mainly aimed at monitoring data of long sequences such as seepage,seepage pressure,and deformation of reservoir dams.A hybrid model based on improved 3-Sigma is proposed,which regulates the safety threshold range by designing outlier factors.The DBSCAN model is integrated to analyze the sequence distribution and density,extract outliers from the sequence data,determine the safety bearing threshold range of the dam,and then calculate the custom warning coefficient to ultimately determine the warning threshold range,thereby guiding the safe operation and management of reservoir dams.The experimental results show that compared to commonly used outlier detection models,the three indicators of precision(99％),F1 score(0.97),and AUC-PR(0.99)all perform well and are feasible in the field of dam safety monitoring.
Dam safety monitoringAnomaly detectionOutlier factorHybrid model
NETL
NSTL
万方数据
