Research On Risk Assessment And Early Warning Of Landslides Triggered By Rainfall Along Electrical Power Transmission Line

Landslides occur frequently along transmission lines,which seriously threaten the safe operation of transmission lines and the power supply of cities.However,in the work of power grid disaster prevention and mitigation in China,the research on geological disaster started late and remains undeveloped.And its ability to deal with extreme meteorological conditions is insufficient.So,it is urgent to carry out the risk assessment and early warning research of geological disaster for the purpose of power grid disaster prevention and mitigation.In view of the characteristics of power grid system,this paper selects the area along the transmission lines across 7 provinces and cities as the research area,and takes raininduced landslide along the transmission line as the main object.Based on rich geological,geographical,environmental and meteorological data,and using different methods such as field investigation,theoretical analysis,GIS spatial analysis,numerical simulation,to carry out deep research of the risk assessment method and key meteorological warning technology of transmission line landslide disaster under rainfall condition.The main contents and achievements are:(1)Through the statistics of the spatial and temporal distribution characteristics of landslides on transmission lines,the main conditioning factors are summarized,and rainfall is determined to be the main inducing factor.(2)Based on the historical rainfall data and landslide inventory,the correlation between rainfall and landslide in the study area was analyzed.Considering geographical,geological,meteorological and environmental factors,the study area is divided into 17sub-regions.The rainfall return period and the rainfall threshold of transmission line landslide in each region are obtained through statistical analysis.Through the inversion analysis of extreme rainfall events,it is found that nearly 50% of landslides are induced by extreme rainfall events.(3)The basic geological data,meteorological and hydrological data,relevant geographical information data and historical landslide inventory of the study area were collected and analyzed,and the key conditioning factors for landslide susceptibility analysis were extracted;Based on the historical landslide inventory and landslide conditioning factors,statistical model and machine learning model were used to evaluate the susceptibility of different spatial scales:(1)nine factors such as lithology,elevation,slope were selected,and based on the information value model the susceptibility maps of the transmission network of 7 provinces and cities were carried out.The evaluation results showed that landslide mainly occurred in eastern Sichuan basin,low mountainous area in western Hubei,east Midwest of Hunan,and Zhejiang.(2)Using two mapping units,grid unit and slope unit,the landslide susceptibility analysis of Wanzhou-Badong key section transmission corridor is carried out.Firstly,15 initial influencing factors were selected,and the factors were tested and selected by variance inflation factor and information gain ratio methods.It was found that topographic factors and environmental factors had a high degree of influence on landslide development.Based on logistic regression and random forest model,the susceptibility models of key research areas were constructed.From the analysis of the model accuracy,it can be seen that the susceptibility model and the result obtained by the random forest model are ideal.Slope unit can significantly improve the display effect of landslide susceptibility mapping of transmission lines,but it is difficult to combine with machine learning model due to the small sample size.(4)The landslide risk assessment model based on conditional probability is established.The temporal probability of the landslide risk is expressed as the combination of the occurrence of characteristic rainfall conditions,and the probability of landslide induced by such rainfall conditions.The hazard is expressed by the conditioning probability of the two aspects.It is found that the calculation of exceedance probability based on landslide threshold can better express the hazard.The quantitative evaluation model of vulnerability is established from two aspects: the disaster resistance ability of transmission tower and the magnitude of landslide.The risk assessment results of transmission line landslide are obtained by combining hazard and vulnerability.(5)Taking Yanzi landslide in Badong County as an example,the risk assessment method of typical single landslide along transmission line was studied.Based on the Gumbel distribution,the extreme rainfall in Badong County under different return periods was calculated,and six different rainfall scenarios were selected.Monte Carlo method was used to calculate the landslide failure probability and landslide surface displacement,and the surface deformation was transformed into the deformation of transmission tower foundation.Based on the distribution of root deformation of transmission tower foundation and surface displacement of landslide,the vulnerability of bearing element and landslide magnitude were established.And the vulnerability curve of transmission pole tower under landslide was also established.The failure probability is used to represent the risk of single landslide,and the quantitative calculation results of economic risk of landslide are obtained based on the analysis of vulnerability and economic value.By comparing the risk calculation results of different rainfall scenarios,it is found that the indirect risk caused by transmission tower damage,such as power outages,is much greater than the direct risk caused by tower damage itself.(6)Based on historical landslide sample data and historical rainfall data,the shortterm warning method of landslide meteorological risk of transmission lines are established;Based on B/S architecture,a landslide meteorological warning system for transmission lines is built,which consists of three modules: early warning criterion construction,GIS spatial analysis and Web publishing.The system realizes the real-time warning function.The accuracy of early warning is verified from two aspects of historical rainfall early warning inversion and landslide field investigation.