Research And Implementation Of Data Fusion Methods For Multi-Source Heterogeneous Data In EMU's Whole Life Cycle

With the continuous development of high-speed railway in China,the demand of high-quality EMU is also increasing.In order to guarantee the safe operation of EMUs by more advanced means,it is important to analyze the EMU's life cycle data.However,under the current situation,EMU data is distributed among multiple business systems in different units,and there are many problems such as different data structures,inconsistent patterns and inconsistent standards.Therefore,in view of the present situation that the multi-source and heterogeneous data of the EMU is difficult to be managed in a unified way,this paper studies how to integrate the data fusion technology with the EMU data management needs,and transform the data distributed in different business systems into the unified format,accurate,component-oriented data,which is conducive to the EMU life cycle of the integrated data management.The main work of this paper includes the following points:(1)This paper analyzes the storage status of multi-source and heterogeneous data of EMU in detail,summarizes the problems that need to be solved when data fusion,and uses data extraction and data cleaning technology to pretreat the EMU multi-source heterogeneous data,making it structure unified and easy to data fusion.(2)Based on the analysis of data fusion related algorithms,ontology theory,data extraction and cleaning theory,a multi-source and heterogeneous data fusion system of EMU is proposed.Two key technologies,Pattern Matching and Data Conflict Resolution,are researched.This paper proposed a method of pattern matching based on EMU ontology knowledge base and composition similarity and a data conflict resolution method based on Markov logic network and combined rules,which are validated by the real life cycle data of EMU.Experiments show that the above two methods can effectively solve pattern matching and data conflict resolution problems in EMU data fusion process.(3)In this paper,the proposed data fusion system is applied to the real life cycle EMU data management scene,realizing the conversion,integration and unity of the data from the car-oriented level to the component level.At last,this paper takes the bogie as an example to show some data organization view...