Train Operation Prediction Using Vehicle Load Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automatic train operation systems fail to accurately predict future operating conditions due to variations in crowding across train vehicles, particularly near stations with high passenger activity, leading to potential delays and operational inefficiencies.
Innovation Solution
An automatic train operation assistance device that collects and analyzes operation data, including load factors, to extract features such as travel and stop times, and constructs a prediction model using machine learning to forecast future train positions and operating conditions, accounting for variations in passenger loading.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical processing based on past stop time is used to predict train operation, then prediction can be performed, but accuracy is insufficient when crowding varies across vehicles
Solution Approach 1:
The patent segments the train into individual vehicles and further segments each vehicle into multiple zones (front, middle, rear). Load sensors are distributed across these zones to capture localized crowding conditions. This segmentation allows the system to account for spatial variations in passenger distribution, improving prediction accuracy without requiring overly complex centralized monitoring.
Solution Approach 2:
The patent introduces load sensors as intermediary devices that indirectly measure passenger crowding conditions. These sensors detect weight changes in each vehicle zone, providing data about passenger boarding and alighting patterns. This intermediary measurement approach enables accurate prediction of stop duration without directly tracking individual passenger movements, balancing precision with system simplicity.
2Device complexity
If uniform crowding assumption is made across all vehicles, then prediction model is simple, but accuracy deteriorates when vehicles near stairs or ticket gates have high crowding
Solution Approach 1:
The patent applies local quality by assigning different prediction weights to different vehicle zones based on their crowding characteristics. Vehicles or zones near stairs, ticket gates, or platform ends are identified as having higher crowding potential and are given greater weight in the stop time prediction calculation. This localized approach improves accuracy by focusing computational resources on the most impactful areas rather than treating all vehicles uniformly.
Solution Approach 2:
The patent dynamically changes the prediction parameters based on real-time load sensor data. When certain vehicles show abnormal crowding patterns (e.g., sudden weight increases indicating passenger boarding), the system adjusts the predicted stop duration for those vehicles accordingly. This parameter adaptation allows the model to respond to varying crowding conditions without requiring a completely complex restructuring of the prediction framework.
3Measurement precision
If detailed load factor data for each vehicle is collected, then prediction accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent implements partial action by selectively processing load data from only the most critical vehicle zones for prediction purposes. Rather than uniformly processing all sensor data from every vehicle, the system identifies and prioritizes data from zones with higher impact on stop duration (such as vehicles at platform ends or near ticket gates). This selective processing maintains high prediction accuracy while reducing overall computational burden.
Solution Approach 2:
The patent merges load data from multiple zones within the same vehicle to create aggregated vehicle-level metrics. By combining data from front, middle, and rear zones, the system creates a comprehensive view of each vehicle's crowding state without needing to process every individual zone separately in all calculations. This merging strategy reduces data processing complexity while preserving the essential information needed for accurate predictions.
Data Source
AI summary
An automatic train operation assistance device includes a data extraction unit that acquires operation data including the operation of a train and load factors of the vehicles of the train from a data collection device that collects equipment data on the train, a feature extraction unit that extracts, using the operation data, operation data on an inter-station including a travel time taken by the train to travel from a first station to a second station and a stoppage time for which the train is stopped at the first station or the second station, as features, and an operation prediction unit that constructs a prediction model using the features, and predicts the future operating conditions of the train including the future position of the train, using the prediction model and current features.