Risk Assessment Of Water Distribution Network Pipeline Failures Based On Spatial Aggregation Attributes And XGBoost Algorithm

Pipe failure is an important issue affecting the safety of water distribution network.Frequent pipe failure infect a lot to people’s production and life,the risk assessment of water distribution network failure has gradually become an important means of scientific management of water supply network,which aims to help water companies arrange maintenance and maintenance plans scientifically and rationally,solve the potential hidden danger of pipe failure.According to the existing research,the properties of soil mechanics and physical and chemical indicators,traffic load,construction quality,pipe material,pipe age,flow,pressure,are all influencing factors for the occurrence of pipe failure,of which pipe material,pipe age,flow rate,pressure are clear,easy to quantify,and easy to apply in the mathematical model.Most of the remaining attribute factors are often difficult to accurately apply in the quantitative application of the model due to data acquisition and quantitative description uncertainty.In fact,most of these attributes have spatial continuity,and similar pipelines have similar influencing factors for pipe failure.The spatial aggregation of the pipe failure reflects the influence of the surrounding environment on it,but in the process of pipe failure risk assessment,these features are often not effectively used,resulting in unsatisfactory risk assessment results.In this paper,the XGBoost algorithm is used to construct a pipe failure risk assessment model,and the impact of the pipe failure space aggregation on the model risk assessment is studied.The main results obtained in the study are as follows:(1)The spatial aggregation of the location of the pipe failure is analyzed,and the spatial aggregation parameter "Dist" is constructed based on this,to replace the factors which are difficult to collect and difficult to quantify.The chi-square test proves that the spatial distribution of pipe failure does not obey homogeneous Poisson distribution,which proves that the burst pipe is not randomly distributed in space.Using the Moran’s I index to analyze the spatial autocorrelation of the pipe,the results show that the p-value is less than 10"5,and the probability of the squib’s random distribution in space is extremely low;Moran’s I>0,which proves that the burst tubes are distributed in a spatially aggregated state.The spatial aggregation parameter "Dist" is obtained by reconstructing the simple and easy-to-collect data such as the number of bursts and the spatial distance,which reflects the influence of environmental factors on the burst(2)Using XGBoost algorithm to establish a risk assessment model for blasting pipes and it was proved that the spatial aggregation parameter "Dist" can effectively improve the accuracy of the risk assessment model.The XGBoost algorithm was used to establish the risk assessment model of the Pipe failure,and resampling was used to improve the unbalanced data set;and the grid search was used to adjust the parameters to improve the performance of the model.Appending the spatial aggregation parameter "Dist" as one of the influencing factors into the explosion risk assessment model effectively improves the accuracy of the explosion risk assessment model.The average AUC of the 10-fold cross-validation cross-validation of the bursting risk assessment model is as high as 0.864,and the AUC of the 10 cross-validations are all above 0.850,which proves that the XGBoost algorithm based on the spatial aggregation attribute has high accuracy and good stability.(3)Constructing pipeline maintenance priority index combining impact consequence index and explosion risk assessment to provide technical support for a scientific and rational arrangement of maintenance and repair plans.A pipe explosion consequence index is constructed based on the flow of the pipe explosion and its connected pipes and the pipe diameter of the pipe explosion.The pipeline maintenance priority index combined with the pipe explosion risk index and pipe explosion consequence index is prioritized The pipeline maintenance priority index can formulate a maintenance plan for water supply enterprises,give priority to the maintenance and repair of high-priority pipelines,allocate the maintenance resources scientifically and reasonably,and provide objective technical support.