Research On The Evaluation Method Of Soil Liquefaction Possibility Based On Chinese Code Parameters

Engineering construction has developed rapidly in our country,the construction of Resilience Urban has also become the target of disaster prevention and mitigation.As an importance link in disaster prevention and mitigation,soil liquefaction has importance engineering value.However,our current liquefaction evaluation method recommended by the code was formed 40 years ago,can't meet current project.Foreign methods are not in line with our engineering habits in the part of ground motion parameters,cannot be well implemented in our country.The development of practical liquefaction evaluation method suited to our national conditions should be one of the research target.In terms of engineering practice,the less parameters were selected the effect better.Using only the most important parameters and the simplest way and under the condition of ensuring engineering accuracy,complete the evaluation of liquefaction possibility under these conditions should be the aim of research.At the same time,the results also help to improve the understanding of the nature and influencing factors of soil liquefaction.With the rapid development of machine learning and the accumulation of liquefaction survey samples,it makes it possible to evaluate the liquefaction possibility of soil layer with the most important parameters and the simplest method.However,the research results in this field are still limited.The aim of this paper is to develop new liquefaction evaluation method which can contribute to our engineering construction.The practical is mainly manifested in three points:Firstly,the evaluation parameter selection match with our existing methods and consistent with our existing engineering habits;The second is to develop existing deterministic method into probability methods;The third is to find out the main parameters and the simplest way to form the soil liquefaction probability evaluation.Based on the above research purpose,this paper focuses on the following several aspects:1.Based on existing international public data from earthquake survey,build the mearsured shear wave velocity sample library soil liquefaction and standard test sample library.With the two sample library as the input,using Logistic Regression method,two calculation methods of liquefaction probability was established based on a_max and Vs or Nm.In the meantime,the critical value is given under the formula of specified probability level,this can provide the support engineering use for our country related specification revision.2.Based on machine learning,the order of importance on liquefied parameters were studied.On this basis,the liquefaction possibility of double parameter based on a_max&V_s,a_max&N_m were built.The rule is easy,the discriminant accuracy rate can get 80%.This research provides a very simple method for liquefaction evaluation.3.On the basis of liquefied parameters importance order,provide the Nm threshold method based on Gini index.In other words,this method can evaluate the soil liquefaction by the threshold standard test N_th easily.In different design acceleration range,this paper puts forward the corresponding threshold N_th,and the reliability of the threshold standard N_th is tested.4.Indentifed the site have V_s and N_m date at the same time,and analyzed these double parameters typical site sample.Compared the above methods of V_s and N_m with the corresponding methods by NCEER.The consistency of the above methods is studied.