Reactive Railway Scheduling via Network Partitioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current railway planning and scheduling systems are inadequate for large-scale networks with unconstrained numbers of trains, stations, and track lines, as they fail to dynamically react to continuous traffic delays, deviations, and disruptions, leading to suboptimal throughput and transit times.
Innovation Solution
A system that continuously executes sense and respond cycles to regenerate reactive online train schedules by interactively partitioning the railway network into sub-networks, using processors to analyze data, reroute trains, and resolve conflicts, thereby optimizing train schedules across the network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional automated planning methods are used, then planning accuracy is improved, but the system cannot handle large-scale networks with unconstrained numbers of trains, stations, and track lines
Solution Approach 1:
The patent divides the large railway network into multiple manageable sub-networks or zones, each handled by dedicated scheduling agents. This segmentation allows the system to maintain planning accuracy for individual segments while scaling to handle the overall large network, resolving the contradiction between precision and network scale capability.
Solution Approach 2:
The patent introduces a hierarchical dimension to scheduling, with different levels of scheduling agents operating at different granularities. Higher-level agents handle strategic planning across the entire network, while lower-level agents manage tactical scheduling in specific zones, enabling the system to accommodate large network scales without sacrificing planning accuracy.
2Adaptability or versatility
If human dispatchers manually monitor and control trains, then flexibility in handling disruptions is improved, but the system becomes intractable for large networks with thousands of trains
Solution Approach 1:
The patent implements autonomous scheduling agents that independently monitor their local sub-networks and make real-time scheduling decisions without human intervention. These agents self-adjust to disruptions and continuously optimize their schedules, providing the flexibility needed for disruption response while avoiding the complexity of manual control for large networks.
Solution Approach 2:
The system incorporates continuous feedback loops where scheduling agents receive real-time information about train status, disruptions, and network conditions, then automatically adjust schedules accordingly. This feedback mechanism enables flexible disruption response while keeping the system complexity manageable through automated decision-making rather than manual control.
3Speed
If dispatchers use local heuristics for train scheduling, then local decision-making speed is improved, but network-wide performance deteriorates due to lack of holistic consideration
Solution Approach 1:
The patent segments the network into multiple sub-networks with dedicated scheduling agents, each capable of rapid local decision-making. Meanwhile, coordination mechanisms ensure that these local decisions align with network-wide optimization goals, achieving both fast local response and improved overall throughput.
Solution Approach 2:
The system merges the speed advantage of local heuristics with the holistic perspective of network-wide optimization by implementing multi-level scheduling agents. Local agents make rapid decisions within their zones while continuously coordinating with higher-level agents that consider network-wide effects, combining the benefits of speed and comprehensiveness.
4Reliability
If the system continuously monitors and re-plans all trains in real-time, then responsiveness to disruptions is improved, but computational complexity becomes uncontrollable
Solution Approach 1:
The patent segments the computational workload by dividing the network into sub-networks handled by separate scheduling agents. Each agent continuously monitors and re-plans only within its local scope, maintaining high disruption responsiveness while keeping individual computational tasks manageable and preventing overall complexity from becoming uncontrollable.
Solution Approach 2:
Instead of re-planning the entire network continuously, the system applies partial re-planning only to affected sub-networks when disruptions occur. This partial action approach maintains reliable disruption responsiveness while significantly reducing computational complexity compared to full-network re-planning.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed is method and system for continuously re-generating reactive on-line train schedules for trains running in a large railway network. Railway network partitioned based on user configuration, into first type comprising trunk line and feeder line sub-networks, and second type comprising supervisory dispatch control territories. Sense and respond cycle is continuously executed on multi-processor computing environment, senses dynamic data from field about train movements, and other changes from users. For each first type sub-network, degree of deviation is computed from incumbent plans and congestion in sub-networks. Using degree of deviation and congestion, trains are rerouted and suitable scheduling methods are chosen for each sub-network and executed in parallel and first level train schedules are sent to second level train schedulers working on second type sub-networks which in parallel identify and resolve conflicts among first level train schedules. Second level train schedules are collated to generate reactive on-line network train schedule.