Research On CBR Decision System For Railway-landslide

Building the decision system of railway line design scheme in landslide area based on CBR(Case-based reasoning),aims to provide supports for the decisions(avoidance or passing with harnessing)of railway line selection in geological disasters area.The mass design and consulting datas in the past can be organized together properly by CBR technology,then applied to decision-making through modifying and reusing,which can reduce the waste of resources and improve work efficiency,and also can overcome the difficulty of knowledge acquisition that expert system has.This paper attempts to build a railway-landslide decision system based on CBR,according to CBR’s 4R cycle(Retrieval,Reuse,Revise,Retrain),and carries out research on methods of each module,and tests its effect according to the actual case.The main contents and basic conclusions of this study are:1.The influence factors of railway-landslide decision-making have been discussed and summarized,the index system of influence factors has been built and quantified.The difficulties of the decision-making of railway-landslide have been sought,and the feasibility of applying CBR to railway line scheme decision-making in landslide area has been researched.Based on the decision-making process of railway route choosing in landslide area and the technical route of the CBR decision system,the framework of the railway-landslide CBR decision system has been constructed,then tested by practical application,the final conclusion is the same with the actual result.2.To build the railway-landslide CBR decision-making system,the detailed realization method of case representation,case retrieval,case reuse and case maintenance has been studied according to the CBR’s 4R cycle.As for case-representation module,Microsoft Access has been used to establish relational database for railway-landslide cases,and the SQL language has been used to add,query,generate report in this database as examples of operation,which have verified the operability of this relational database used to represent the railway-landslide cases;As for case-maintenance module,information entropy and rough set have been used as two different ways to reduct attributes,a calculation example has been carried out,the consistent results have verified the correctness and effectiveness of those two methods of reduction;As for case-reuse module,the artificial neural network technology has been used to reuse the cases that are similar to the target cases and also used as countermeasure when case matching fails.An calculation example has been given to recommend the use of neural network models under different conditions.3.Research on the construction of database for railway-landslide cases has been carried out,the relational database of railway-landslide cases has been designed,and in detail,the entity-relationship model has been designed,the relational tables have been created and standardized,and the relationships between relational tables has been built.Finally the database has been put in use as example.OLAP(On Line Analytical Processing)report has been established according to the simulation of actual demand,and in an actual scheme comparison problem,the database has been used to filter the cases,both of them come to one conclusion:the application requirements have been achieved through the relational database.4.Case analysis:the CBR decision system has been applied to a railway line scheme comparison problem in landslide area,and the case analysis has been carried out,the decision result which is the same with the actual result has verified the system’s practicability and effectiveness.Overall,it’s feasible to use CBR to construct Railway-landslide decision system,which has certain theoretical value.And it’s feasible to apply the CBR decision system based on the framework of two-library structure to the problem of railway line scheme decision-making in landslide area.In addition,using artificial neural network to deal with the failure of case matching is effective and feasible.