Research On Data Mining Algorithm Of On-Line Electromagnetic Rail Defect In Cloud Computing Environment

In recent years,China's railway has entered a rapid development stage.With the increase of rail transport capacity,the corresponding number of trains,and the train running speed,higher requirements are put forward to the railway operation safety issues.One of the important means to ensure traffic safety is the rail inspection work.At present,China's railway rail inspection work mainly relies on the combination of artificial flaw detection car and large flaw detection mode.However,this mode of operation can only be run in the window period of the train operation,limited by the train running time limit and low efficiency,and not adapt to current high-speed,intensive train operation.The development of on-board electromagnetic rail detection method makes the rapid testing possible.By installing online electromagnetic rail detection equipment in the train,can improve the speed and efficiency of testing without occupying tracking time.Under this background,the methods of data analysis and processing,such as large data and cloud computing,was proposed to deal with the data produced by vehicle-mounted online electromagnetic rail testing device,and a data management platform was developed to solve the real-time error and visualization of flaw detection data,improving the efficiency of flaw detection.The main work in this paper is as follows:Firstly,the research background and research significance of rail detection is introduced,as well as the current domestic and international development of rail inspection equipment.On this basis,the research significance of the flaw data analysis and the current situation at home and abroad,as well as the development prospect of the cloud computing in the field are introduced.And the importance of the data mining analysis of the online electromagnetic rail detection in the cloud computing environment is clarified.Secondly,the cloud data analysis platform based on B/S architecture for cloud storage data on cloud servers was built,which allows data analysis and display in the cloud computing environment.First the two types of database which are sequential and relational respectively were used to reduce the database access pressure when the data volume is too large.Then the front-end interface was designed to achieve the platform user login,rights management,password recovery and other basic functions,especially the graphical display of the flaw detection data.At last a robust java language was used to establish logical frame,connecting front-end pages and background database.Thirdly,the data of rail detection is analyzed.The original data are divided into original types first,and then the data preprocessing steps such as normalization and filtering denoising are carried out to obtain high quality data.According to the clustering and classification algorithm,the classification of the damage type,such as crack type,triangular pit type and internal damage,is classified according to the characteristics of the data.Finally,damages of the same type are classified by clustering and classification algorithm to identify the damage level,and the performance of each algorithm is compared according to the classification effect.