Railway Track Collapse Monitoring System in Mining Area based on KALMAN Filter

Abstract: Facing on the phenomenon of railway track collapse in mining areas, Basing on KALMAN filter, a kind of railway track collapse monitor system was proposed in this study. Firstly, the basic design principle of the system was analyzed. Secondly, MEMS gyroscope and GPS module were used as attitude measurement components, adopting PLC to collecting real-timing sampling of gyroscope and GPS information, customizing Windows CE operating system of ARM to control the whole system. What is more, the data of gyroscope and GPS were fusion by using KALMAN filter method on MATLAB software. The test results showed that the system has accurate readings in the flat and bumpy road, Besides, the system can feedback the results effectively when the angle error is more than 0.4 degrees. The system conformed to the index of railroad track dynamic detection in China, which can be widely used in railway track subsidence monitoring.


Introduction
Railway track transportation has an irreplaceable role in our daily life.Track collapse may cause the train run unstably and create serious accidents [1].In remote areas, ordinary artificial static test was low efficiency and the period of calling large track inspection vehicle was too long, which could not meet the requirement of the safety production.What is more, multiple sensor components and complex algorithm in most detection system cannot apply for the track collapse test [2,3].This article designed a low cost portable track collapse monitoring system and related algorithms.In the end, the system experiment was proposed to verify the effect.
2 The hardware and software design of the system

The Overall Structure of the System
Track collapse monitoring system consists of three parts, acquisition unit, storage unit and data process unit.Firstly, system collected angles and corresponding location information through MEMS gyroscope and GPS module by means of PLC.Secondly, PLC sent the test data to the Win CE system ARM through serial port.What is more, KALMAN algorithm was used to make the data fusion and remove noise [4,5] on personal computers.Finally, the analysis data of track collapse situation were shown on screen.Figure 1 is the overall structure of the system.

Hardware design
The accuracy of system depends on the performance of gyroscope sensor and GPS module [6].System chose LPMS-CU gyroscope and TIANBAO BD970 GPS module after thinking of comprehensive factors.Latitude and longitude information that GPS transfer back can be directly collected by PLC [7].It is no need to have signal amplification, which only need to demodulation for location information in the PLC program.But the gyroscope signal is relatively weak.So it is important to design signal amplification circuit [8] on gyroscope.Circuit diagram designed was shown in Figure 2. KALMAN filter algorithm (optimization regression data processing algorithm) obtained optimal value angle by fusing the measured value of GPS and gyroscope.If we want to estimate the actual angle values of k time, we need to predict the angle of k time according to the angle value of k-1 time [10,11].And then system calculated the Gaussian noise according to the predicted angle of k time [12].Finally, the KALMAN filtering algorithm continually did recursive variance to the end.The formulas were derived according to the method above. ( ( / ( ( 1) ) The formulas above were updated equation of KALMAN filter.A and B were state transition matrix.K was KALMAN gain.Z(k) was the observation matrix and H was parameter matrix.We performed the KALMAN filtering algorithm through MATLAB software on PC terminal System, which can eliminate the gyroscope drift and noise reduce noise, so as to optimize angle value through KALMAN filtering algorithm.
3 System experiment

Experimental method
Experiment Chose flat and bump roads instead of the railway track and equipped sensors on the car.Car speed ranges from 1m/s to 1.5 m/s, Both running distance were ten meters and sampling period were 0.2 seconds.Three experiments were carried out in respectively three days.Each test was able to detect both direction Angle and location information.Finally, drawing the conclusion by using KALMAN filter algorithm on the computer.

Experimental result
1. Firstly, dedicated level protractor was used to measured the static values of sampling points before the system test.Three measurements made the average to determine the scope of the system angle adjustment.2. Afterwards, setting sample frequency of 5 Hz after the system run smoothly.Three experiments were carried out in respectively three days to get the measuring values of corresponding position coordinates.
3. Finally, data was uploaded to the computer and system obtained the absolute error values of corresponding position coordinates(the difference between readings after filtering and static values) and covariance value of the system by means of KALMAN algorithm.Covariance value determines the system precision.
The smaller covariance value is, the higher system precision will be.The absolute errors value determines whether the collapse is out of limits.In case of the measured horizontal Angles, the typical experimental data of horizontal Angle in flat and bumpy roads were showed in the Table 1 and Table 2.

Conclusion
This paper designed a set of railway track subsidence monitoring system based on MEMS gyroscope.System design, hardware selection circuit and software algorithm were accomplished in the study.Through the actual system test and the analysis of relevant data, the conclusion is made as follows: 1) System mainly obtained gyroscope Angle and GPS location signals by PLC.GPS and gyroscope data were fusion by KALMAN filtering algorithm, which have a good dynamic response in the experiment.2) The system test results showed that the system has accurate readings in the flat and bumpy road.Besides, the system can feedback the results effectively when the angle error is more than 0.4 degrees, The system conforms to the index of railroad track dynamic detection in China, which can be widely used in railway track subsidence monitoring.
great deal of effort and energy to give me a lot of advice and guidance.Afterwards, Thanks to my family, thanks to my lab partner that gave me help in the process of the experiment.It is because of your help and support, I can overcome all difficulties and doubt until the smooth completion of this paper.Finally, I would like to thank to the science and technology support plan funded project of Jiang Su province.

Figure 1 .
Figure 1.The overall structure of the system

Table 1 .
Horizontal angle measured changes in the flat road test

Table 2 .
Horizontal angle measured changes in the bumpy road test Note: covariance value is 0.0048 in the flat road and 0.0036 in the bumpy road.