Research And Application Of Dynamic Inspection System For Track Longwave Irregularities Based On Vision/Inertial Fusion

Track dynamic inspection technology is an important means to realize track geometry status evaluation and operation and maintenance in China.This technology combines linear structured light and inertial measurement methods to measure track irregularity,and has the characteristics of high speed and high precision.As the speed of the train continues to increase,the long-wave irregularity of the track begins to affect the driving comfort,which needs to be measured and managed;the demand for deformation monitoring of bridges,settlements and other sections is also increasing.However,the existing track inspection system has the shortcomings of poor accuracy and high cut-off speed in measuring long-wave irregularity,which limits the application of measurement data in maintenance and deformation monitoring.The realization and application of high-precision measurement and application of track long-wave irregularity at sub-millimeter level in high-dynamic and high-noise driving environment is the focus of this paper.The main research work includes the following aspects:(1)The wavelength range of track long-wave irregularity management and detection has not yet been strictly defined.Analysis of the influence of long-wave irregularity will help to define the needs of dynamic detection.Firstly,the characteristics of typical longwave irregularity diseases are discussed according to the generation mechanism;then combined with vehicle dynamics simulation and measured vehicle body dynamic response,it is found that 100?200m is the main sensitive wavelength range under highspeed driving conditions,which needs to be detected and managed.(2)Aiming at the problem that the existing system is difficult to accurately measure the long-wave irregularity,the paper clarifies the main error sources of the system through theoretical derivation and experimental analysis: 1)The "error angle of attack" between track and the IMU's sensitive axis,the profile and alignment measurement are greatly affected by the pitch and yaw angle respectively;2)the random noise of the inertial sensor;3)the measurement error of the roll angle of the inertial sensor carrier;4)the harmful components of the uncompensated sensor.The measurement error is greatly affected by the vibration state of the train,and it exhibits low-frequency characteristics after integration.(3)In order to reduce the negative influence of "error attack angle" on the measurement of long-wave irregularity,a dynamic measurement model based on "angular displacement" is proposed.The model uses a gyroscope and a linear structured light ranging sensor to observe the "angular displacement" change of the short vector distance difference on the orbital plane,and then derives an improved angular velocity measurement method.Aiming at the problem of high frequency attenuation in the measurement results,a fusion method based on complementary filtering to achieve fullband irregularity measurement is proposed.The vehicle dynamics model simulation example shows that compared with the traditional acceleration measurement method,the accuracy of the complementary filtering measurement method is improved by 24% to80%,and it has the advantages of less influence by the detection speed and low noise sensitivity.(4)Aiming at the problems that the measurement accuracy of the existing system is not stable enough and the applicability of the measurement results is insufficient,a related optimization method is proposed.Firstly,in view of the constant drift of the profile in the curve section,a correction method for the harmful component of centripetal acceleration is proposed.Secondly,aiming at the problem of the deterioration of the accuracy of cant measurement under variable speed conditions,a bogie attitude determination method based on error Kalman filtering is designed,and the zero-speed corrected attitude observation model is optimized.In the tests,the accuracy of cant measurement is increased by 89%;Then,in order to increase the applicability of the detection data in the field maintenance,the general dynamic measurement calculation formula of mid-chord offset method.Finally,in order to solve the problem of missed inspection of track geometric parameters at low speed,a dynamic measurement method based on angular velocity is proposed,which realizes the "zero-speed detection" of short-wave irregularity.(5)A prototype system for dynamic detection of long-wave irregularity based on Comprehensive Inspection Train is designed and optimized.In the dynamic experiment,the measurement accuracy is systematically evaluated by the internal and external coincidence accuracy.The dynamic measurement of track irregularity with a cut-off wavelength of 200 m cut-off and a Mid-Chord value of 60 m is completed under complex driving conditions.The measurement repeatability of thealignment and profile is 0.5 mm and 0.8 mm,compared with the existing comprehensive detection.In the test,it is found that the severe high-frequency vibration at a high speed of 350 km/h reduces the reproducibility of the measurement method,and the vibration isolation method still needs to be continuously researched and improved.(6)The rail inspection system currently lacks an effective error assessment method.Therefore,a set of spatial correlation dynamic measurement error propagation evaluation method is established for the application scenario of orbit dynamic measurement.Only considering the influence of random noise of the sensor,the system is regarded as a linear time-invariant system excited by white noise,and the transfer relationship between the error source and the measurement error is calculated;and the inertial reference measurement SNR concept is proposed,and the quantitative analysis of measurement error demonstrates the feasibility of achieving sub-millimeter relative measurement accuracy under different cut-off wavelengths,sensor random noise coefficients,and driving speeds,and in-depth analysis of the quantitative impact of various random error sources on the measurement accuracy of track geometric parameters.The selection of inertial sensors and the setting of cut-off speed in the design provide a theoretical basis.It provides systematic and quantitative theoretical guidance for the design of future orbit detection systems.