Refine
Year of publication
Document Type
- ZIB-Report (139)
- In Proceedings (84)
- Article (63)
- Book chapter (5)
- In Collection (3)
- Other (3)
- Doctoral Thesis (2)
- Book (1)
- Habilitation (1)
- Master's Thesis (1)
Keywords
- line planning (13)
- Optimierung (9)
- integer programming (9)
- column generation (7)
- combinatorial optimization (6)
- Periodic timetabling (5)
- public transport (5)
- Column Generation (4)
- Integer Programming (4)
- Linienplanung (4)
Institute
- Mathematical Optimization (240)
- Mathematics of Transportation and Logistics (106)
- Network Optimization (48)
- ZIB Allgemein (15)
- Modeling and Simulation of Complex Processes (8)
- AI in Society, Science, and Technology (4)
- Mathematics of Health Care (3)
- Computational Medicine (2)
- Mathematical Algorithmic Intelligence (2)
- Mathematics for Life and Materials Science (2)
We study a complex planning and scheduling problem arising from the build-up process of air cargo pallets and containers, collectively referred to as unit load devices (ULD), in which ULDs must be assigned to workstations for loading. Since air freight usually becomes available gradually along the planning horizon, ULD build-ups must be scheduled neither too early to avoid underutilizing ULD capacity, nor too late to avoid resource conflicts with other flights. Whenever possible, ULDs should be built up in batches, thereby giving ground handlers more freedom to rearrange cargo and utilize the ULD's capacity efficiently. The resulting scheduling problem has an intricate cost function and produces large time-expanded models, especially for longer planning horizons. We propose a logic-based Benders decomposition approach that assigns batches to time intervals and workstations in the master problem, while the actual schedule is decided in a subproblem. By choosing appropriate intervals, the subproblem becomes a feasibility problem that decomposes over the workstations. Additionally, the similarity of many batches is exploited by a strengthening procedure for no-good cuts. We benchmark our approach against a time-expanded MIP formulation from the literature on a publicly available data set. It solves 15% more instances to optimality and decreases run times by more than 50% in the geometric mean. This improvement is especially pronounced for longer planning horizons of up to one week, where the Benders approach solves over 50% instances more than the baseline
We study the solution of the rolling stock rotation problem with predictive maintenance (RSRP-PdM) by an iterative refinement approach that is based on a state-expanded event-graph. In this graph, the states are parameters of a failure distribution, and paths correspond to vehicle rotations with associated health state approximations. An optimal set of paths including maintenance can be computed by solving an integer linear program. Afterwards, the graph is refined and the procedure repeated. An associated linear program gives rise to a lower bound that can be used to determine the solution quality. Computational results for six instances derived from real-world timetables of a German railway company are presented. The results show the effectiveness of the approach and the quality of the solutions.
An Iterative Refinement Approach for the Rolling Stock Rotation Problem with Predictive Maintenance
(2024)
The rolling stock rotation problem with predictive maintenance (RSRP-PdM) involves the assignment of trips to a fleet of vehicles with integrated maintenance scheduling based on the predicted failure probability of the vehicles. These probabilities are determined by the health states of the vehicles, which are considered to be random variables distributed by a parameterized family of probability distribution functions. During the operation of the trips, the corresponding parameters get updated. In this article, we present a dual solution approach for RSRP-PdM and generalize a linear programming based lower bound for this problem to families of probability distribution functions with more than one parameter. For this purpose, we define a rounding function that allows for a consistent underestimation of the parameters and model the problem by a state-expanded event-graph in which the possible states are restricted to a discrete set. This induces a flow problem that is solved by an integer linear program. We show that the iterative refinement of the underlying discretization leads to solutions that converge from below to an optimal solution of the original instance. Thus, the linear relaxation of the considered integer linear program results in a lower bound for RSRP-PdM. Finally, we report on the results of computational experiments conducted on a library of test instances.
The covering of a graph with (possibly disjoint) connected subgraphs is a funda-mental problem in graph theory. In this paper, we study a version to cover a graph’svertices by connected subgraphs subject to lower and upper weight bounds, and pro-pose a column generation approach to dynamically generate feasible and promisingsubgraphs. Our focus is on the solution of the pricing problem which turns out to bea variant of the NP-hard Maximum Weight Connected Subgraph Problem. We com-pare different formulations to handle connectivity, and find that a single-commodityflow formulation performs best. This is notable since the respective literature seemsto have widely dismissed this formulation. We improve it to a new coarse-to-fine flowformulation that is theoretically and computationally superior, especially for largeinstances with many vertices of degree 2 like highway networks, where it provides aspeed-up factor of 5 over the non-flow-based formulations. We also propose a pre-processing method that exploits a median property of weight-constrained subgraphs,a primal heuristic, and a local search heuristic. In an extensive computational studywe evaluate the presented connectivity formulations on different classes of instances,and demonstrate the effectiveness of the proposed enhancements. Their speed-upsessentially multiply to an overall factor of well over 10. Overall, our approach allowsthe reliable solution of instances with several hundreds of vertices in a few min-utes. These findings are further corroborated in a comparison to existing districtingmodels on a set of test instances from the literature
Flight planning, the computation of optimal routes in view of flight time and fuel consumption under given weather conditions, is traditionally done by finding globally shortest paths in a predefined airway network. Free flight trajectories, not restricted to a network, have the potential to reduce the costs significantly, and can be computed using locally convergent continuous optimal control methods.
Hybrid methods that start with a discrete global search and refine with a fast continuous local optimization combine the best properties of both approaches, but rely on a good switchover, which requires error estimates for discrete paths relative to continuous trajectories.
Based on vertex density and local complete connectivity, we derive localized and a priori bounds for the flight time of discrete paths relative to the optimal continuous trajectory, and illustrate their properties on a set of benchmark problems. It turns out that localization improves the error bound by four orders of magnitude, but still leaves ample opportunities for tighter bounds using a posteriori error estimators.
Globally optimal free flight trajectory optimization can be achieved with a combination of discrete and continuous optimization. A key requirement is that Newton's method for continuous optimization converges in a sufficiently large neighborhood around a minimizer. We show in this paper that, under certain assumptions, this is the case.
The ongoing electrification of logistics systems and vehicle fleets increases the complexity of associated vehicle routing or scheduling problems.
Battery-powered vehicles have to be scheduled to recharge in-service, and the relationship between charging time and replenished driving range is non-linear.
In order to access the powerful toolkit offered by mixed-integer and linear programming techniques, this battery behavior has to be linearized.
Moreover, as electric fleets grow, power draw peaks have to be avoided to save on electricity costs or to adhere to hard grid capacity limits, such that it becomes desirable to keep recharge rates dynamic.
We suggest a novel linearization approach of battery charging behavior for vehicle scheduling problems, in which the recharge rates are optimization variables and not model parameters.
The currently most popular approach to handle non-linear battery behavior for electric vehicle scheduling is to use a linear spline interpolation of the charge curve. We show that this can lead to approximate models that underestimate the charge duration and overestimate the state of charge, which is not desirable. While the error is of second order with respect to the interpolation step size, the associated mixed-integer linear programs do not scale well with the number of spline segments. It is therefore recommendable to use coarse interpolation grids adapted to the curvature of the charge curve, and to include sufficient safety margins to ensure solutions of approximate models remain feasible subjected to the exact charge curve.
Convergence Properties of Newton's Method for Globally Optimal Free Flight Trajectory Optimization
(2023)
The algorithmic efficiency of Newton-based methods for Free Flight Trajectory Optimization is heavily influenced by the size of the domain of convergence. We provide numerical evidence that the convergence radius is much larger in practice than what the theoretical worst case bounds suggest. The algorithm can be further improved by a convergence-enhancing domain decomposition.