Rail Vehicle Motion Estimation Device Using AI Plausibility Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Estimating movement values such as speed or acceleration for rail vehicles is complex in poor adhesion conditions due to the influence of anti-slip and anti-skid regulations, which affect measured axle speeds, leading to uncertain or error-prone results.
Innovation Solution
A two-stage estimating device comprising a trained artificial intelligence system that calculates movement values based on measured values, followed by a correction device that checks the plausibility of these values and corrects them if necessary, using axle speed values and wheel grip signals to ensure accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a trained artificial intelligence system is used to calculate movement values based on measured values, then the estimation capability is improved, but the reliability may be uncertain under poor adhesion conditions
Solution Approach 1:
The estimation device is segmented into two independent stages: a trained AI system for calculation and a correction device for validation. This segmentation allows each component to specialize - the AI system focuses on processing complex measured values while the correction device focuses on reliability verification, resolving the contradiction between estimation capability and reliability.
Solution Approach 2:
The correction device acts as an intermediary between the AI system and the final output. It receives the reference value from the AI system, performs plausibility checks using additional measured values, and either outputs the reference value or generates a corrected movement value. This intermediary layer ensures reliability without compromising the AI system's estimation capability.
2Reliability
If anti-skid and anti-skid controls are applied to handle poor adhesion conditions, then the safety is improved, but the complexity of the overall system increases
Solution Approach 1:
The correction device performs self-service by automatically detecting implausible reference values and correcting them using measured values from the environment. It monitors the plausibility of AI-generated estimates and self-corrects without external intervention, maintaining safety while avoiding the need for complex external control systems.
Solution Approach 2:
The correction device implements a feedback mechanism where the plausibility of the reference value is continuously checked against measured values. When implausibility is detected, the system feeds back correction actions by generating corrected movement values, creating a closed-loop control that ensures safety without excessive complexity.
3Reliability
If the correction device performs comprehensive plausibility checks on all reference values, then the reliability is improved, but the processing time increases
Solution Approach 1:
The correction device applies partial action by performing plausibility checks only when necessary - specifically when the reference value falls outside expected ranges based on measured values. Instead of checking every single reference value exhaustively, the system applies corrections selectively based on detected implausibility, reducing processing time while maintaining reliability.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The invention relates, inter alia, to an estimation device (10) for estimating a motion value (Vfinal, afinal) describing a movement parameter of a rail vehicle (500). According to the invention, the estimation device (10) comprises a trained artificial intelligence system (11) that calculates a reference value (Vki, aki) based on measured values (M), and a correction device (12) is arranged downstream of the trained system (11). This correction device checks the reference value (Vki, aki) and outputs it as the motion value (Vfinal, afinal) of the estimation device (10) if the reference value (Vki, aki) passes a plausibility check, and corrects it by generating the motion value (Vfinal, afinal) of the estimation device (10) if the reference value (Vki, aki) fails the plausibility check.