Research On Data Reduction And Discretization Of Metro Rigid Catenary Based On Rough Set

With the rapid development of China's economy and the continuous advancement of urbanization,people's demand for urban travel mode is getting higher and higher,and urban metro emerges as the times require.Rigid catenary has been widely used in urban metro due to its simple structure,high reliability and small space occupation.In order to realize fault management of rigid catenary,data mining and analysis of fault detection information is particularly important.However,the rigid catenary fault information and detection data collected directly from the scene often have many disadvantages such as inconsistent recording mode,non-uniform format,large scale,noise-containing data and continuity.If such a low-quality data set is directly applied to data mining,the desired mining effect will not be achieved.Therefore,the data preprocessing of fault information and detection data of rigid catenary has very important research value and practical significance.Rough set theory can effectively process and analyze inaccurate,inconsistent and incomplete information and data.In this paper,the application of rough set theory in fault information and detection data of metro rigid catenary is studied.The main work is as follows:Based on the rough set model based on equivalence relation,the basic concepts of equivalence relation,equivalence class,exact set,rough set,upper and lower approximation,kernel,reduction and discretization are studied,and the basic framework of rough set theory is discussed.According to the basic theory of information coding,the types,items,attributes and fault types of rigid catenary components are reasonably classified and unified coding is designed.Based on Power Designer and Oracle,the database structure of rigid catenary components and faults is designed,the conceptual model and physical model of the database are built,and the database of rigid catenary components and faults is created.The relative attribute reduction algorithm based on greedy strategy and the reduction algorithm based on discernibility matrix are studied.These two reduction algorithms are used to optimize the types and items of rigid catenary components in Rose2.The two reduction algorithms are compared and evaluated in terms of reduction results,heuristic information,algorithm steps and reduction time.Equal-width method,equal-frequency method,K-Means clustering method and discretization algorithm based on Boolean logic are studied.Equal-width method,equal-frequency method and K-Means clustering method are implemented in Python.Discretization algorithm based on Boolean logic is implemented in Rosetta.Using the above four discretization algorithms,the height,pull-out value and pantograph-catenary contact pressure of the station detection data in Fenghuang Xincun-Shayuan area of Guangzhou Metro Line 8 are discretized.