| With the rapid development of Chinese transportation industry,the mileage of railway operations has increased rapidly,and high-speed railway has become a national business card.The foundation of the safe,reliable and smooth operation of trains is a stable subgrade.The deformation of subgrade will lead to a decrease in the stability of train operation and bring certain hidden dangers to safety.In severe cases,it may even cause the rail to break and the train to overturn.Therefore,it is very important to accurately monitor the state of subgrade.In order to overcome the shortcomings of traditional measurement methods and the inapplicability of new measurement methods,this dissertation applies machine vision measurement technology to subgrade settlement monitoring,and studies the image-based high-speed railway subgrade settlement monitoring method covering the entire road section,with high precision,automation and strong real-time performance.In this dissertation,a monitoring terminal is designed and implemented according to this method.Aiming at the difficulty of determining maintenance time of the monitoring terminal,the evaluation of system state is transformed into the evaluation of image quality of spots,and a method of evaluating the spot quality combining subjective and objective is proposed.In order to further improve the measurement accuracy,a data adjustment optimization strategy is proposed.The main research contents are as follows:Firstly,the principle of image-based high-speed railway subgrade settlement measurement is studied,and a single-stage monitoring method is proposed to realize the settlement value measurement of one monitoring points.When the single-stage monitoring systems are connected end-to-end,the mathematical model of multi-stage subgrade settlement monitoring system with chain network is established.On the basis of analyzing the requirements of this monitoring system,the structural design of monitoring system is completed.In this dissertation,the monitoring system is divided into 7 main modules,and the hardware selection is completed according to the module functions.The monitoring terminal using STM32 chip as a main controller,which cooperates with the image processing module and the communication module,completes the subgrade settlement value measurement and the settlement data transmission.In terms of software,image preprocessing and spot center positioning algorithm based on circle fitting are studied,the workflow of monitoring terminal and image processing module is introduced in detail,and a C#-based Web API program is designed to receive and store measurement data on the user side.The monitoring terminal model is designed by Solid Works software,and then the model is installed and tested.The monitoring terminal described in this dissertation can measure subgrade settlement with a resolution of 0.143 mm.Then,in order to solve the problem that the system maintenance time is difficult to determine in a complex environment,the monitoring terminal state evaluation is transformed into the spot image quality evaluation,on the basis of analyzing the monitoring method.A light spot quality evaluation model combining subjective and objective is proposed.The effectiveness of this algorithm to evaluate spot image quality is verified by simulation experiments and field experiments,and the threshold value of spot image quality is determined to be 0.95.Finally,in order to improve the measurement accuracy of the monitoring system,a round-trip measurement adjustment strategy applied to adjacent monitoring stations is proposed.After applying this method,the measurement accuracy can be improved by 2times.Aiming at the problem of error accumulation in the chain structure,an adjustment optimization strategy of constructing a closed network is proposed,and this adjustment is completed with closed conditions.The effectiveness of this data adjustment strategy to improve the measurement accuracy is verified by simulation experiments. |
PDF Full Text Request