Research On The Intelligent Pre-control Of Metro Construction Safety Risks Oriented To Design Process

With the rapid expansion of urban city scale and population in China,the conflicts between the diversified travel demands and the limited traffic resources have become more obvious.Considering advantages of the metro in high-speed,large capacity,low pollution and high efficiency,it has been a back bone of the public transportation system in many cities.China is undergoing a boom period of urban metro construction in recent years.Whilst the metro construction safety accidents occur frequently,which cause massive casualties as well as economic losses.Therefore,it is significant to study on how to promote metro construction safety and avoid metro construction accident.Most studies have study on the potential of design to improve construction safety performance.However,few literatures focus on the theory,model and pre-control methods for reducing safety risk of metro construction from the perspective of design.Therefore,from the perspective of the design process and based on the theory and practice orientation,the meanings of the safety design of metro construction were defined,the formation mechanisms were explored,and systematic researches were conducted from two aspects of intelligent prediction of construction safety risk and intelligent pre-control methods.The results findings have contributed to making the scientific and reasonable preventive measures in metro construction so as to provide the theoretical and technical methods.(1)On the basis of the detailed literature review,the overseas and domestic research status of design for safety,safety risk factors of metro construction,safety risk monitoring and assessment approaches,the prediction and early-warning intelligent inference for safety management in metro construction were summarized,comments were made,and the shortages of the existing researches were pointed out.And then these shortages and the problems in the practice,the aims,contents and research framework of this study were specified.(2)The conception of design for safety(DFS)in metro construction was defined.Based on the analysis of the construction safety management standards,norms and literature analysis,twenty factors related to safety management in the whole process of metro construction were screened.Based on the integrated interpretative structural modeling and MICMAC(a French term:Matriced'Impacts Croises-Multiplication Appliance a un Classement,MICMAC)methods,a hierarchical structure and a driving dependence graph which were used to interpret how these design factors impact the metro construction were generated.The relationships between design and construction safety accidents were analyzed from accident causation sequence theory and track crossing theory,and its formation mechanism was analyzed.(3)The types and characteristics of 246 of metro construction accidents in China from 2002 to 2016 were analyzed in order to define the research scope.According to the formation mechanism of design for safety in metro construction,this study screened twenty-one critical influence factors in five dimensions: hydrogeology,engineering geology,ambient environment,engineering design plan and construction scheme design.Based on the data gathered and subsequently analyzed from the questionnaire surveys of experts in academia and engineering,the availability and validity of factors were verified.(4)Combining the complementary advantages of Rough Sets(RS)and Support Vector Machine(SVM),this paper constructed a metro construction safety risk intelligent prediction model based on RS-SVM.Without changing the quality classification of the sample,the RS method was used to reduce the redundant indicators through genetic algorithm.As a result,the sample of instability failure of foundation pit was reduced from 21 to 12 key factors,and the sample of surroundings damage was reduced 21 to 9 key factors.Through eliminating redundant indicators,redundant indicators could be eliminated,the dimensions of sample data were simplified,and the training efficiency and accuracy of the intelligent prediction model were improved.(5)According to the characteristics of metro construction safety risk,a binary tree support vector machines was designed,and the particle swarm optimization(PSO)was proposed to optimize the parameters of SVM model.A PSO-SVM intelligent prediction model was built for the safety risk intelligent prediction.Then,the PSO-SVM intelligent prediction model was trained and tested with the sample data of eliminated redundant indicators.The results showed that the PSO-SVM intelligent prediction model was confirmed to be consistent with the actual situation of training set and test set.In addition,the PSO-SVM intelligent prediction model could analyze the complex coupling effects between design factors and construction safety risk,and provide valuable information for design factors,which could promote safety performance and decrease the safety risk of metro construction.