A Lagrangian Heuristic for Robustness, with an Application to Train Timetabling
نویسندگان
چکیده
Finding robust yet efficient solutions to optimization problems is a major practical issue that received large attention in recent years. Starting with stochastic programming, many of the approaches to robustness lead to a significant change in the problem formulation with respect to the non-robust (nominal) case. Besides requiring a much larger computational effort, this often results into major changes of the associated software. Lagrangian heuristics form a wide family of methods that work well in finding efficient (i.e., low cost) solutions for many problems. These methods approximately solve a relaxation of the problem at hand through an iterative Lagrangian optimization scheme, and apply several times a basic heuristic driven by the Lagrangian dual information (typically, the current Lagrangian costs) so as to hopefully update the current best feasible solution. In this context, the underlying Lagrangian optimization iterative method has the main purpose of producing “increasingly reliable” Lagrangian costs, while diversifying the search in the last iterations, when the Lagrangian bound is very close to convergence. The purpose of this paper is to propose, for the first time to the best of our knowledge, a very simple modification of the Lagrangian optimization scheme capable of dealing with robustness. This modification is based on two simple features: (a) we modify the problem formulation by introducing artificial parameters intended to “control” the solution robustness; and (b) during Lagrangian optimization, we dynamically change the weight of the control parameters so as to produce subproblems where robustness becomes more and more important. In this way, during the process we can easily collect a set of, roughly speaking, “Pareto optimal” heuristic solutions that have a different tradeoff between robustness and efficiency, and leave the final user the choice of the ones to analyze in more details—e.g., through a time-consuming validation tool. As a proof-of-concept, our approach is applied to the well-known aperiodic Train Timetabling Problem (TTP) on a corridor, and is computationally analyzed on real-world test cases from the Italian Railways, showing that it produces within much shorter computing times solutions whose quality is comparable with those produced by existing approaches to robustness. This proves the effectiveness for the specific application, suggesting that a simple modification of existing Lagrangian heuristics is a very promising way to deal with robustness in many other cases.
منابع مشابه
A discrete-event optimization framework for mixed-speed train timetabling problem
Railway scheduling is a complex task of rail operators that involves the generation of a conflict-free train timetable. This paper presents a discrete-event simulation-based optimization approach for solving the train timetabling problem to minimize total weighted unplanned stop time in a hybrid single and double track railway networks. The designed simulation model is used as a platform for ge...
متن کاملA Lagrangian heuristic algorithm for a real-world train timetabling problem
The train timetabling problem (TTP) aims at determining an optimal timetable for a set of trains which does not violate track capacities and satisfies some operational constraints. In this paper, we describe the design of a train timetabling system that takes into account several additional constraints that arise in real-world applications. In particular, we address the following issues: • Manu...
متن کاملDemand-oriented timetable design for urban rail transit under stochastic demand
In the context of public transportation system, improving the service quality and robustness through minimizing the average passengers waiting time is a real challenge. This study provides robust stochastic programming models for train timetabling problem in urban rail transit systems. The objective is minimization of the weighted summation of the expected cost of passenger waiting time, its va...
متن کاملCut generation for an employee timetabling problem
Motivated by an industrial application, we study a specific employee timetabling problem. Several investigations are being conducted: a lower bound by Lagrangian relaxation, a heuristic based on a cut generation process and an exact method by Benders decomposition. Experimental results on real and generated instances are reported.
متن کاملTrain Timetabling on double track and multiple station capacity railway with
Train scheduling is one of the significant issues in the railway industry in recent years since it has an important role in efficacy of railway infrastructure. In this paper, the timetabling problem of a multiple tracked railway network is discussed. More specifically, a general model is presented here in which a set of operational and safety requirements is considered. The model handles the tr...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Transportation Science
دوره 46 شماره
صفحات -
تاریخ انتشار 2012