TL;DR
This paper addresses the critical need for high-accuracy track irregularity measurements in modern railways. It proposes and validates a motion-constrained GNSS/INS integration approach that achieves millimeter-level relative spatial accuracy, crucial for ensuring train safety and passenger comfort, particularly for shortwave track irregularities.
What Problem Does This Paper Address?
Modern high-speed railways demand precise track geometry measurements for safety and comfort. While GNSS/INS integration offers a promising solution for efficient track surveying, questions remain about its ability to achieve the centimeter-to-millimeter level accuracy required for detecting subtle track irregularities. Existing methods often focus on absolute accuracy, which doesn't fully capture the nuances of relative positional variations critical for identifying track deformations. This paper aims to bridge this gap by defining and assessing the relative spatial accuracy required for shortwave track irregularity measurements and demonstrating how a motion-constrained GNSS/INS system can meet these stringent demands.
Methods
The core of this research lies in defining and evaluating relative spatial accuracy. The authors introduce metrics like Mean Square Error (MSE) and Allan Deviation (ADEV) to quantify this accuracy, focusing on the variation of navigation errors over specific distances rather than absolute deviations from a true path.
A key methodological contribution is the development and application of motion-constrained GNSS/INS integration. Recognizing that railway trolleys have inherent kinematic constraints (e.g., wheels staying on rails), the authors incorporate these Non-Holonomic Constraints (NHCs) into the INS mechanization. This approach leverages the predictable motion of the trolley to enhance the accuracy of the integrated navigation solution, particularly when GNSS signals might be weak or intermittent. The integration algorithm uses a 21-state error model, accounting for position, velocity, and attitude errors, as well as IMU sensor biases and scale factors.
Datasets
The study utilizes data collected from a field test conducted on a ballastless track section of the second Lanzhou-Xinjiang high-speed railway in China. The test track was approximately 1.0 km long and situated in an open-sky environment. The data was collected using a navigation-grade LINS812 GNSS/INS system mounted on a track geometry measurement trolley. Reference data was obtained using a high-precision digital level (Trimble DiNi) and an Amberg Slab Track GRP1000 system.
Key Insights and Ideas
- Relative Accuracy is Paramount: For track irregularity measurement, the precise relative positions of points along the track are far more critical than their absolute positions in a global frame. Small variations in relative position directly translate to track deformation.
- Motion Constraints Unlock Millimeter Accuracy: Standard GNSS/INS struggles with the required accuracy due to limited dynamics and potential signal issues. Incorporating the non-holonomic constraints of the railway trolley significantly improves the system's ability to achieve the necessary millimeter-level relative spatial accuracy.
- ADEV and MSE as Key Indicators: Allan Deviation and Mean Square Error are effective tools for evaluating the stability and random error components of the navigation solution, directly correlating with the quality of track irregularity measurements.
- Shortwave vs. Longwave: The paper specifically focuses on shortwave irregularities, which require higher spatial resolution and thus more stringent relative accuracy requirements.
Key Results
The study demonstrates that:
- Standard GNSS/INS: In its standard mode, the GNSS/INS integration achieved absolute vertical position errors with an RMS of around 2.35-2.64 mm and lateral position errors with an RMS of around 2.31-2.79 mm. However, the MSE at a 5m distance was 0.79-0.83 mm (vertical) and 0.91-0.92 mm (lateral), and ADEV was 0.49-0.54 mm (vertical) and 0.57-0.59 mm (lateral). These values were larger than the derived thresholds for shortwave track irregularities (0.67 mm for MSE, 0.16 mm for ADEV).
- Motion-Constrained GNSS/INS: With motion constraints applied, the system showed significant improvements. The absolute vertical position errors had an RMS of approximately 0.41-0.51 mm. Crucially, the MSE at a 5m distance dropped to 0.31 mm (vertical) and 0.51 mm (lateral), and the ADEV dropped to 0.13-0.14 mm (vertical) and 0.25 mm (lateral).
- Meeting Requirements: The motion-constrained GNSS/INS system successfully met the derived accuracy requirements for shortwave vertical irregularities (MSE < 0.67 mm, ADEV < 0.16 mm). For lateral irregularities, the MSE met the threshold, while ADEV was slightly higher but still significantly improved.
Limitations and Future Work
- Limited Test Track Length: The field tests were conducted on a relatively short track section (approximately 1 km). While sufficient for shortwave analysis, longer-term performance and longwave irregularity assessment might require testing on more extensive track segments.
- Open-Sky Environment: The tests were performed in an open-sky environment, ensuring good GNSS signal availability. Future work should investigate the system's performance in scenarios with GNSS signal outages (e.g., tunnels).
- Human-Powered Trolley: The trolley was pushed by human force at a moderate speed (3 m/s). While this is suitable for demonstrating the core principles, real-world applications might involve higher speeds and automated propulsion, which could introduce different dynamic characteristics.
- Future Directions: The authors propose incorporating low-cost sensors like odometers and laser scanners to enhance the system's capability in scenarios with poor or no GNSS signals, further broadening its applicability.
Summary updated