Train Weight Estimation via Rail Vibration and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for monitoring railway components are limited by the complexity of data collection, particularly in accurately measuring properties like train weight, which requires correlation with ground truth data, thereby restricting their applicability.
Innovation Solution
A system comprising sensors, processing, storing, and analyzing components that sample and process sensor data relevant to trains, using machine learning techniques and digital twin technology to estimate properties like speed, weight, and wheel health, and associate these with interactions between railway components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sensor methods are used to measure train properties like weight, then measurement precision can be achieved, but device complexity and difficulty of detecting and measuring increase due to requiring ground truth correlation
Solution Approach 1:
The patent replaces traditional mechanical sensor systems that require ground truth correlation with a machine learning-based system that uses acceleration patterns and vibration data to infer train properties. The analyzing component processes sensor data through trained models to estimate weight, speed, and other characteristics without requiring direct mechanical measurement infrastructure.
Solution Approach 2:
The patent introduces machine learning models as intermediaries between raw sensor data and train property measurements. These models act as mediators that translate acceleration patterns and vibration signals into accurate estimates of train weight and other properties, eliminating the need for direct correlation with ground truth data from complex measurement infrastructure.
2Measurement precision
If traditional sensor methods are used to measure train properties, then measurement precision can be achieved, but the difficulty of detecting and measuring increases due to requiring ground truth correlation
Solution Approach 1:
The patent replaces traditional mechanical measurement methods requiring ground truth correlation with machine learning-based inference systems. The analyzing component uses trained models to directly estimate train weight from acceleration and vibration patterns, eliminating the complex process of correlating sensor data with ground truth measurements.
Solution Approach 2:
The machine learning models are trained offline on historical data and then autonomously perform measurement tasks without requiring ongoing ground truth correlation. Once trained, the system serves itself by automatically inferring train properties from sensor data without human intervention or complex measurement infrastructure.
3Device complexity
If machine learning techniques are used to estimate train properties, then device complexity is reduced, but measurement precision may be compromised without ground truth correlation
Solution Approach 1:
The patent performs preliminary training of machine learning models using ground truth data during an offline phase. This preliminary action allows the models to learn accurate relationships between sensor data and train properties before deployment, ensuring measurement precision is maintained while simplifying the operational system complexity.
Solution Approach 2:
The patent creates digital twin models that replicate the physical train system's behavior. These virtual copies are trained to match real train characteristics and can accurately estimate properties from sensor data, maintaining measurement precision while reducing the need for complex physical measurement infrastructure.
4Reliability
If comprehensive sensor data collection is implemented to monitor all train properties, then reliability of monitoring is improved, but device complexity and loss of time for data processing increase
Solution Approach 1:
The patent extracts and focuses on the most informative sensor signals (acceleration and vibration patterns) that contain sufficient information to infer multiple train properties. By selecting only the critical sensors and features, the system maintains high monitoring reliability while reducing overall device complexity and data processing requirements.
Solution Approach 2:
The machine learning models are designed to infer multiple train properties (weight, speed, acceleration patterns) from a single set of sensor measurements. This multi-functional approach allows comprehensive monitoring with minimal sensors, improving reliability without increasing device complexity or data processing burden.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system effectively monitors and forecasts the health status of railway components, enabling accurate property estimation and predictive maintenance, thereby enhancing operational safety and efficiency.
Implementation Method 1
such sensors may be used to obtain acceleration patterns, which may be induced by a passing train and measured, for instance, by an accelerometer affixed to the rail or sleeper
Implementation Method 2
Acceleration patterns may contain information about properties of the train that induced said vibrations
Data Source
AI summary
The present invention discloses a system and a method for monitoring properties of at least one train is disclosed. The system comprises at least one sensor component configured to sample at least one sensor data relevant to the at least one train. The system further comprises at least one processing component configured to process the at least one sensor data. The system comprises at least one storing component configured to store the at least one sensor data, and at least one analyzing component.