Public Transit Schedule Optimization via Parallel Local Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current public transportation scheduling methods fail to efficiently minimize waiting times at transfers, especially in complex systems with multiple transportation modes and scenarios, as they often result in large-scale optimization problems that standard solvers cannot handle effectively.
Innovation Solution
A quadratic mixed-integer program formulation is introduced to optimize public transportation schedules, allowing for parallel evaluation of waiting time impacts and a closed-form solution for transfer choices, enabling an efficient parallel local search algorithm to reduce waiting times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data-driven approaches with two-stage stochastic linear programs are used to model transfer waiting times, then measurement precision of waiting times is improved, but device complexity and computational burden increase significantly
Solution Approach 1:
The patent segments the large-scale stochastic optimization problem into multiple smaller subproblems by decomposing the network into components and handling each scenario separately. This allows the complex problem to be solved in manageable pieces while maintaining data-driven accuracy.
Solution Approach 2:
The patent uses heuristic methods that provide approximate solutions rather than exact optimal solutions, trading off some precision for computational efficiency. The heuristics generate sufficiently good solutions quickly without requiring expensive, time-consuming exact optimization algorithms.
2Manufacturing precision
If standard optimization solvers are used to solve large-scale stochastic programs, then manufacturing precision of schedules is improved, but productivity decreases due to computational intractability
Solution Approach 1:
The patent divides the large-scale optimization problem into smaller subproblems that can be solved independently and in parallel. By segmenting the network and scenarios, the computational burden is reduced while maintaining solution quality through coordinated optimization of the subproblems.
Solution Approach 2:
The patent employs heuristic methods that provide approximate solutions rather than exhaustive optimal solutions. This partial action approach generates good enough schedules quickly, sacrificing some precision for significant gains in computational speed and productivity.
3Ease of operation
If expert knowledge and rules of thumbs are incorporated into timetable design, then ease of operation is improved, but measurement precision of waiting times deteriorates
Solution Approach 1:
The patent uses data-driven models that incorporate actual system performance feedback to refine and improve timetable designs. By using real transfer waiting time data from the transportation system, the methodology objectively quantifies performance and enables continuous improvement based on measured results rather than subjective expert judgment.
Data Source
AI summary
Methods and systems are disclosed for optimizing public transportation schedules. Shifts from current schedules associated with a public transportation system can be evaluated. The impact on waiting times can then be determined, based on evaluating the shifts from the current schedules. Schedules associated with the public transportation system are then optimized based on the impact on the waiting times. Public transport schedules can thus be optimized by minimizing the waiting time during, for example, transfers. Using ticket validation data to construct a realistic scenario-based model of the waiting times, a goal of this approach is to compute shifts of the current schedules that reduce the overall expected waiting time.