Detection and Correction of Abnormal Data in Water Distribution Networks Based on LOF Model
The measurement errors in flow and pressure data often interfere the real-time hydraulic model EPANET-RTX for water transmission pipelines.To address this issue,the Local Outlier factor(LOF)algorithm is used to detect abnormal samples,then the Z-score algorithm is used to detect abnormal parameters,and the k-neighborhood mean method is used to correct abnormal parameters,based on a large amount of measurement data.This approach achieves fast and accurate high-dimensional vector detection and correction,thereby ensures EPANET-RTX stably and reliably tracking the real-time hydraulic status of water transmission pipelines.Taking the JS water transmission pipeline network as an example,abnormal data detection and correction are carried out.The accuracy rate of abnormal data recognition reaches 99.53％,both the quality and the speed of correction data meet the needs of real-time hydraulic models.
Water transmission pipelinesAbnormal data detectionLOF algorithmZ scoreK-neighborhood mean method
NETL
NSTL
万方数据
