Dam Seepage Safety Assessment And Prediction Research Based On Set Pair Cloud Model

The total amount of water resources in China is small,the per capita is insufficient,and the distribution of time,space and area is extremely uneven,which can not meet the needs of water shortage areas or seasons.The construction of embankments and the regulation of natural water quantity by reservoirs are the practical needs for human beings to prevent drought and flood disasters and reasonably develop water resources,and play an active role in agricultural irrigation,urban water supply,hydropower development and shipping improvement.In view of this,the number and types of dam foundations in China are large,including some dangerous dams.At the same time,seepage is one of the key factors for safe operation of various types of dams,and comprehensive evaluation of seepage safety is of great significance to dam safety.On the other hand,dam seepage safety assessment has some shortcomings and limitations in the establishment of grade standards,construction of index system and index weighting.Most of the existing studies are based on analysis of time-consuming data,but no research has been carried out on seepage behavior prediction.Based on this,this paper takes a concrete face rockfill dam as an example and proceeds from its multi-source monitoring data to carry out the dam seepage safety evaluation and prediction research considering the shortcomings in the current seepage safety evaluation field.Firstly,on the basis of existing research,an evaluation index system of dam seepage safety is established by means of frequency analysis,reliability analysis and effectiveness analysis,and a combination weighting model of subjective and objective is established by combining game theory and two methods.Then,starting from the characteristics of dam seepage and its index,five-level fuzziness criteria are obtained by using digital characteristics and cloud model theory.In addition,after introducing sparrow algorithm and improving its shortcomings,the parameters of bi-directional long-term and short-term neural network are optimized,and a dam seepage pressure prediction model is built.Finally,by introducing set pair analysis and cloud model theory,a seepage safety evaluation model is established and applied to an example analysis of a concrete face rockfill dam.The results show that:(1)Based on literature research,reliability analysis and effectiveness analysis,a multiperspective and multi-method screening method for dam seepage safety evaluation indexes is established,and an evaluation index system including rainfall,water level,temperature,seepage pressure value and seepage pressure change rate is established,and the non-linear correlation between seepage pressure value and rainfall,water level and temperature is proved.Based on the long-sequence monitoring data of an example project,the boundary level criteria based on the digital characteristics of mean and standard deviation are established.Then,with the help of cloud model theory,five-level fuzziness level criteria based on cloud model theory are put forward,which fuzzifies the level criteria and weakens the threshold concept.(2)The index scaling method is introduced to improve the intuitionistic fuzzy AHP and optimize the subjective weighting result of seepage index.At the same time,objective weights of each index are determined by using correlation coefficient method,which further combines game theory method for combination weighting and brings into play the advantages of the two weighting methods.Finally,the results of calculation by combination weighting method and traditional weighting method are compared.The results show that the weighting method of seepage safety evaluation index proposed in this paper is more scientific and reasonable,and can provide reference for the weighting link in the field of seepage safety evaluation.(3)Long-term and short-term neural network prediction model coupled with multistrategy improved sparrow algorithm is introduced to predict seepage pressure of a certain face rockfill dam.The results show that the convergence speed and optimization accuracy of the sparrow algorithm are improved.On this basis,the improved sparrow algorithm coupled with bi-directional long-term and short-term neural network osmotic pressure prediction model has a high fitting accuracy of 0.987,which greatly improves the precision compared with the basic osmotic pressure prediction model and reduces the error by about 30%.(4)Using the cloud coupling model proposed in this paper,multi-year time series seepage safety evaluation is carried out for a concrete face rockfill dam and its seepage time-varying law is analyzed.The results show that the safety level of seepage in non-flood season is mostly in the level ?-??,and the level ?-??? oscillation changes during flood season due to the influence of rainfall and seepage pressure.In addition,rainfall has the greatest impact on seepage safety and the minimum temperature.Finally,the improved seepage pressure prediction model and seepage safety evaluation model are combined to make up for the shortage of existing research,to realize the risk level prediction of dam seepage,and to provide new ideas and methods for dam safety monitoring and evaluation.