Train Scheduling with Hybrid Answer Set Programming

Abstract

We present a solution to real-world train scheduling problems, involving routing, scheduling, and optimization, based on Answer Set Programming (ASP). To this end, we pursue a hybrid approach that extends ASP with difference constraints to account for a fine-grained timing. More precisely, we exemplarily show how the hybrid ASP system clingo [DL] can be used to tackle demanding planning-and-scheduling problems. In particular, we investigate how to boost performance by combining distinct ASP solving techniques, such as approximations and heuristics, with preprocessing and encoding techniques for tackling large-scale, real-world train scheduling instances.

In-Station Train Movements Prediction: from Shallow to Deep Multi Scale Models

“…This work is a first step towards improving the prediction of in-station train movements, and more tests (on more stations and for a longer period of time) on a larger period of time and space (once trains will restart to circulate after the COVID pandemic) need to be performed to extensively asses the potentiality of this solution. Moreover, research needs to be conducted to assess the potentiality of this tool to improve the state-of-the-art train dispatching systems [5,6].…”

“…In fact, in order to effectively and efficiently plan the dispatching of the trains inside a station, it is required to predict the time needed to traverse each block section [5] (Figure 1). In particular, we need to be able to predict these movements duration at least 1 hour [4][5][6] in advance in order to being able to optimise the dispatching. For this purpose, at time t, to make a prediction at time t+1h, all the time series representing the behaviour of all trains in a sub portion of the network insisting on the station of interest are available from t−∆ to t. Moreover, additional (side) information is available like the planned itinerary of the trains through the station, if the train stops (and in case the scheduled stop time), its delay, day of the week, working day of holiday, type of train, time slot, weather conditions, etc.…”

“…Unfortunately, these data are seldomly available. Nevertheless, the ability to accurately predict the duration of trains movements is crucial for improving the in-station train dispatching [4][5][6]. Train dispatching (both human-or AI-based), in fact, requires accurately predicting the duration of trains movements to efficiently cope with the increase in capacity demand and with delays.…”

A Planning-Based Approach for In-Station Train Dispatching

“…As future work, we are interested in extending our approach to support maintenance operations and preferences, and to consider different and more complex scenarios. Finally, we would like to consider alternative approaches, such as Answer Set Programming (Abels et al 2020).…”

Constraint Logic Programming for Real-World Test Laboratory Scheduling

“…In the case of hybrid systems, (Balduccini 2011) and (Friedrich et al 2016) employ an ASP and CP hybrid approach to solve industrial machine scheduling problems. An application of ASP combined with difference logic for a real-world train scheduling problem can be found in (Abels et al 2020).…”