Conflict-Based Repair Techniques for Solving Dynamic Scheduling Problems

Abstract

Constraint satisfaction problems (CSP) are increasingly used for solving scheduling problems: many global temporal and resource constraints have been developed. However, dynamic CSP (an extension of the CSP framework where the set of variables or/and constraints evolves throughout computation) have not been used for solving dynamic scheduling problems. Moreover, dynamic scheduling problems are themselves seldom studied. Two classical methods are used to solve such problems are: recomputing a new schedule from scratch each time an event occurs (a quite time consuming technique) or constructing a partial schedule and completing it progressively as time goes by (like in online scheduling problems - this is not compatible with planning purposes). Extending work on dynamic arc-consistency, the use of explanations (a set of constraint justifying solver actions) have been introduced for solving dynamic problems. However, no application of such a technique have been made to scheduling problems. We introduce the integration of explanation capabilities within scheduling-related global constraints and show how using such techniques speeds up the solving of dynamic problems (compared to solving a series of static problems)