Multi-Objective Concurrent Intelligent Optimization Of Railway Alignments,Structures And Large-Scale Auxiliary Construction Projects In Complex Mountainous Regions

Alignment determination is the central work of the entire railway design process,which fundamentally controls the railway’s life-cycle performances on safety,economy,comfort and environmental impacts.With Chinese next-generation railways transform from the eastern plains to the western mountainous regions as well as further extend to foreign countries along the Silk Road Economic Belt,a large number of complex mountain railway projects have been planned and built.For these railways,the proportion of complicated bridges as well as tunnels along the alignment has risen dramatically and the difficulty of infrastructure construction has been significantly increased.Thus,the coordinated layout of railway alignments(hereinafter denoted as “RA”),different structures(“DS”)and large-scale auxiliary construction projects(“LACP”)has become an inevitable requirement of mountain railway design processes.However,the current optimization of RA,DS and LACP is mostly based on a “serial & separation mode”,which overlooks the multi-level coupling relations and collaborative evolution characteristics among RA,DS and LACP.Besides,existing studies in this field largely concentrate on the construction cost related single-objective evaluation,which are essentially separated sub-system optimization studies of RA or LACP with the consideration of limited objectives and constraints.As a result,it is difficult to produce an optimized RA-DS-LACP solution from a comprehensively coordinated perspective under different environmental conditions.To this end,viewed from the railway RA-DS-LACP collaborative coupling system,this dissertation focuses on two major research aspects: multi-objective intelligent optimization model and multi-level intelligent search algorithm,to solve intelligent railway alignment optimization problems in complex mountainous regions.The major research contents and contributions are as follows:(1)In this research,the RA,DS and LACP are treated as an integrated system.The spatial geographic information model,RA sub-system optimization model and LACP sub-system optimization model are built and combined.Unified decision variables,constraints and cost functions are established for the entire system.Thus,a basic optimization model of RA-DS-LACP integration is proposed:(1)The structural characteristics of “multi-level,tight coupling,and dynamic evolution” in the process of RA-DS-LACP collaborative design are systematically analyzed.An overall framework of “determining LACPs based on RA,evaluating RA according to LACPs,and evolving RA-DS-LACP concurrently” for modeling and solving the RA-DS-LACP optimization problem is designed.The model elements of RA and LACP subsystems are studied and the integrated framework of these sub-models are proposed;(2)The spatial organization and analysis method of multi-source heterogeneous geographic information required for alignment design is proposed.The accurate mapping and rapid computation between the RA-DS-LACP’s comprehensive decision variables and their spatial distributions are achieved.Based on the existing design constraints of RA and LACP,an anti-hazard constraint set for RA-DS-LACP simultaneous optimization with respect to collapses,landslides,debris flows and fault zones is established.Afterward,the computation method of RA-DS-LACP maximum construction duration constraint is developed.Finally,the geographic information model,RA model,and LACP model are integrated.The basic coordination optimization model for RA-DS-LACP system is therefore formed.(2)Based on the above model,a novel model is proposed to compute the geological hazards of RA-DS-LACP systems when they traverse rockfalls,landslides and debris flows:(1)Regarding the existing geo-hazard regions(including rockfalls,debris flows and landslides),a RA-DS-LACP hazard analysis framework is proposed based on a weighted ranking method.At this stage,by combing the triggering factor analysis of various geo-hazards and the anti-hazard ability of different structures,a danger evaluation model is constructed by integrating a hazard susceptibility analysis and a structural vulnerability analysis for the RA-DS-LACP coupling system;(2)Regarding the potential threats of unknown geo-hazard regions,the study area is separated into hazard outbreak,buffer and fuzzy regions according to specific division standards.Specifically,the alignment geo-hazard danger in buffer regions is analyzed with an energy slope procedure and the potential hazard threats of alignments in fuzzy regions is assessed using an information value model.Therefore,a comprehensive RA-DS-LACP hazard evaluation model covering “existing + potential” geo-threats is formed.(3)A spatiotemporal seismic risk quantitative evaluation model is developed for intelligent RA optimization.Thus,an engineering economy-geological hazard-seismic risk multi-objective model is built for RA-DS-LACP concurrent optimization:(1)Regarding the ground shaking risks,several seismic performance analysis methods are first developed for different structures based on probabilistic fragility curves.Then,by incorporating the structural damage ratio functions,the spatial seismic direct loss of alignments that caused by ground shaking can be computed.Afterward,for different post-earthquake time stages,an operator to predict railway risks during the seismic crisis period is developed based on an event tree.Meanwhile,the railway revenue losses during the post-disaster recovery period are also evaluated.Thus,the “direct + indirect”,“spatial + temporal” ground shaking risk model of the RA-DS-LACP system is constructed;(2)Regarding the earthquake-induced landslide risks,the probabilistic seismic intensity model is first introduced and combined with the critical intensity value for triggering landslides to determine the potential displacements of earthquake-induced landslides.Then,the landslide displacements are linked with specific structural damage states.Thus,the functions of alignment structural damage states,damage ratios and restorations can be integrated to finally estimate the RA-DS-LACP’s direct(spatial)as well as indirect(temporal)risks that posed by earthquake-induced landslides.(4)To solve the above multi-objective optimization model,an “area-corridor-alignment” multi-level intelligent optimization framework is constructed.First,at the “area-corridor” stage,a RA-DS collaborative search method based on a 3-D distance transform(3D-DT)is proposed to generate railway corridors within in large-scale study areas:(1)The multi-criteria tournament decision-making(MTD)approach is adopted in this research.Then,based on robust optimization(RO)theory,a RO-MTD generalized distance that suitable for multi-objective alignment optimization is defined;(2)An erythrocyte-shaped 3-D neighboring mask is designed for scanning search spaces of the 3D-DT.A horizontal-vertical stepwise search as well as a known-unknown backpropagation deduction is combined as the dynamic search strategy of the large study area;(3)The dynamic handling operators of various constraints and the preliminary design method for LACPs are developed.A fitting method for refining 3D-DT paths into railway trend lines is proposed and,then,railway corridors can be generated with the trend lines as their centerlines;(4)In order to improve the optimization efficiency of the 3D-DT,a “parallel linkage algorithm” based on Open MP library is devised,which significantly accelerate the search speed.(5)Then,at the “corridor-alignment” stage,a RA-DS-LACP multi-objective optimization method is proposed based on the particle swarm optimization(PSO)algorithm:(1)A “horizontal-vertical-integration” stepwise strategy is developed for optimizing 3-D alignments.The stepwise initialization and evolution approaches for alignment particles are respectively designed;(2)A “tunnel shaft-access road-integration” stepwise strategy is conceived for determining LACPs based on alignment alternatives.In this regard,a stepwise PSO-divide & conquer-Dijkstra hybrid solution method is proposed;(3)Based on the Pareto optimization theory,a hybrid multi-objective solver integrating a crowding distance computation and a marginal benefit analysis is proposed,which realizes the resolution and balance of complex conflicts among different objective functions;(4)To enhance the global exploration and local exploitation abilities of the algorithm,the updating rationales of inertia weight,velocity constraint and acceleration coefficients during the PSO evolution process are also improved.Finally,the processes of “evaluating RA according to LACPs” as well as “evolving RA-DS-LACP concurrently” can be realized.Based on the above theoretical achievements,the railway “RA-DS-LACP collaborative optimization software” is developed with the author as the major builder,which has been successfully applied to several important railway design institutes such as the China Railway First Survey and Design Institute Group Co.Ltd,the China Railway Eryuan Engineering Group Co.Ltd and the China Railway Siyuan Survey and Design Group Co.Ltd.Using this software,the effectiveness and improvement of proposed models and methods are demonstrated through seven representative railway cases in complex mountainous regions.This dissertation includes 119 figures,33 tables and 348 references.