Determining Capacity of Shunting Yards by Combining Graph Classification with Local Search

Ven, A. van de; Zhang, Yingqian; Lee, Wan-Jui; Eshuis, Rik; Wilbik, Anna

doi:10.5220/0007398502850293

Cited by 3 publications

(13 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Blue edges represent the order of the movements and green edges indicate which service task is completed first. Activity nodes encode specific service tasks and parking tracks as a subscript [21] Previously proposed methods in graph classification can be subdivided into graph kernel methods and topological methods. Graph kernels measure the similarity between graphs by directly comparing substructures within graphs.…”

Section: Graph Classificationmentioning

confidence: 99%

“…Results show that time-related features are the most import for prediction accuracy in the tested data. Later, [21] uses a DGCNN [23] to extract graph features (node classes) directly from the activity graphs of initial solutions without incurring in manual feature engineering. The results show that the DGCNN can achieve similar performance when compared to [6], suggesting that the DGCNN model is able to extract useful features from shunting plans.…”

Section: Graph Classificationmentioning

confidence: 99%

“…In the context of train shunting, the authors of [19] consider the parking of trains as a complete (reinforcement) learning problem. In [6,21], machine learning methods are used to learn the relationship between initial solutions an a feasibility output from the LS heuristic.…”

Section: Introductionmentioning

confidence: 99%

“…We then use a Deep Graph Convolutional Neural Network (DGCNN) [23] as a feature extractor to train a model that predicts the future feasibility of each precedence graph. We formalise the approach of [21] by including a sequence of seen shunting plans during an LS run as training examples. We show how to combine the predictions with the LS to derive a policy that decides to terminate the LS run based on the sequence of intermediate solutions seen so far.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Data-Driven Policy on Feasibility Determination for the Train Shunting Problem

Costa¹,

Rhuggenaath²,

Zhang³

et al. 2020

Machine Learning and Knowledge Discovery in Databases

Self Cite

View full text Add to dashboard Cite

Parking, matching, scheduling, and routing are common problems in train maintenance. In particular, train units are commonly maintained and cleaned at dedicated shunting yards. The planning problem that results from such situations is referred to as the Train Unit Shunting Problem (TUSP). This problem involves matching arriving train units to service tasks and determining the schedule for departing trains. The TUSP is an important problem as it is used to determine the capacity of shunting yards and arises as a sub-problem of more general scheduling and planning problems. In this paper, we consider the case of the Dutch Railways (NS) TUSP. As the TUSP is complex, NS currently uses a local search (LS) heuristic to determine if an instance of the TUSP has a feasible solution. Given the number of shunting yards and the size of the planning problems, improving the evaluation speed of the LS brings significant computational gain. In this work, we use a machine learning approach that complements the LS and accelerates the search process. We use a Deep Graph Convolutional Neural Network (DGCNN) model to predict the feasibility of solutions obtained during the run of the LS heuristic. We use this model to decide whether to continue or abort the search process. In this way, the computation time is used more efficiently as it is spent on instances that are more likely to be feasible. Using simulations based on real-life instances of the TUSP, we show how our approach improves upon the previous method on prediction accuracy and leads to computational gains for the decision-making process.

show abstract

Section: Graph Classificationmentioning

confidence: 99%

Section: Graph Classificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Data-Driven Policy on Feasibility Determination for the Train Shunting Problem

Costa¹,

Rhuggenaath²,

Zhang³

et al. 2020

Machine Learning and Knowledge Discovery in Databases

Self Cite

View full text Add to dashboard Cite

show abstract

“…This formulation is based on a real-life situation, and the results of that competition made clear that for practical application this problem could only be solved by heuristics (Geiger et al, 2018). Since then these heuristics have been further developed by Van Den Broek (2016), Kremer and Buijs (2016), and Van De Ven et al (2019). However, we cannot estimate the quality of the solutions found by these heuristics.…”

Section: Introductionmentioning

confidence: 99%

Train Unit Shunting and Servicing: a Real-Life Application of Multi-Agent Path Finding

Mulderij¹,

Huisman²,

Tönissen³

et al. 2020

Preprint

View full text Add to dashboard Cite

In between transportation services, trains are parked and maintained at shunting yards. The conflict-free routing of trains to and on these yards and the scheduling of service and maintenance tasks is known as the train unit shunting and service problem. Efficient use of the capacity of these yards is becoming increasingly important, because of increasing numbers of trains without proportional extensions of the yards. Efficiently scheduling maintenance activities is extremely challenging: currently only heuristics succeed in finding solutions to the integrated problem at all. Bounds are needed to determine the quality of these heuristics, and also to support investment decisions on increasing the yard capacity. For this, a complete algorithm for a possibly relaxed problem model is required. We analyze the potential of extending the model for multi-agent path finding to be used for such a relaxation.

show abstract