Research On Load Characteristics And Load Spectrum Of Metro Vehicles Based On Condition Division

As the main system of urban rail transit,the fatigue reliability of metro vehicles is very important.Due to the large traffic volume,complex operation conditions and insufficient maintenance of metro vehicles in China,the international general design codes and standards can no longer ensure the safe operation of vehicles within the design life.It is necessary to study the loading characteristics of vehicle structure under actual operation conditions,especially under the line with insufficient maintenance,and compile the load spectrum of metro vehicle bogie frame for actual metro lines.In this dissertation,the bogie frame load of the subway vehicle can be accurately identified,and a universal automatic working condition division method is proposed.The load characteristics of the bogie frame under each working condition are studied,and the load spectrum of each load is compiled according to the working condition.A load spectrum merging method based flow matrices under different conditions is proposed.This dissertation provides a basis for establishing the design and experimental standards of metro vehicle bogie frame for actual metro lines and promoting the line universality of bogie frame load spectrum.The main research contents of this dissertation are as follows:(1)According to the current standard and the structural form of metro trains bogie frame,the load system of the bogie frame is divided.Based on the load identification method of direct measurement,the decoupling calibration test was carried out,and the bogie frame and the auxiliary structure were used as the force measuring components to realize the continuous and accurate identification of the load.(2)In order to analyze the load characteristics of each working condition,establish the load spectrum of each load division of metro vehicle bogie frame,and promote the line universality of the load spectrum,it is necessary to divide the load data into working conditions.This dissertation proposes a working condition division method based on load characteristics,that is,by analyzing the influence degree of each load by different working conditions,find out the working conditions that have a great influence on each load in turn,and adopt different working condition division methods for different loads.It solves the problems of insufficient accuracy and excessive division of partial load caused by the previous unified working condition division method.(3)This dissertation proposes a universal automatic working condition classification method based on computer vision to automatically divide a large amount of data into working conditions.The theoretical basis and how the method are implemented are detailed,and the accuracy of the division results is evaluated.Compared with the traditional method of selecting data features and classifying them according to thresholds,when dealing with different test data,this method does not need to re-select features,calculate thresholds or write programs.It only needs to establish new data labels and train a network according to the division objectives,so as to realize the accurate division of working conditions and the universality of automatic working condition division method.(4)In this dissertation,the characteristics of each load under typical working conditions are studied,including traction,coasting,braking conditions,and the laws of time-domain signals and statistical characteristics of each load under linear and curvilinear conditions,and the correlation between different loads and working conditions under each working condition is obtained.The influence of the line conditions including the rail corrugation and the wheel polygon on the load is studied.For the track corrugation,the road sections with corrugation damage were confirmed,the corrugation value was predicted based on the quadratic integration method in the frequency domain,and the relationship between the corrugation and the load amplitude was fitted.For the wheel polygon,based on the frequency domain analysis of the stress and load response data and the comparison of the data before and after the wheel,the polygon wear order was confirmed,and the influence of the wheel polygon and orbital wear on the load was compared.(5)Based on the load characteristics,a method for non-calibrated load identification based on the least squares support vector machine algorithm is proposed,and the theory of the least squares support vector machine is introduced.Taking a simple nonlinear model as an example,the identification ability of the least squares support vector machine method to the nonlinear system inverse model is verified.Finally,the vertical load identification of the motor is realized by the randomly selected stress response near the motor.(6)Rain flow counting and grouping calculation are carried out on the measured load data under different working conditions,and the load spectrum of each load division is compiled.The characteristics of load spectrum under different working conditions are compared and analyzed.In order to promote the line universality of the load spectrum and predict the load spectrum of other road sections,a load spectrum merging method based on the rain flow matrix is proposed.The load spectrum merging method based on the operating mileage ratio of each working condition is implemented,and the accuracy of the load spectrum prediction results is verified.Based on a certain type of metro vehicle,this dissertation puts forward a method system from load identification to working condition division,working out load spectrum under different working conditions,and combining load spectrum under various working conditions for load prediction,which promotes the research on the line universality of load spectrum.There are 68 figures,20 tables and 80 references.