CONSTRAINT-DIRECTED SEARCH: A CASE STUDY OF JOB-SHOP SCHEDULING

This thesis investigates the problem of constraint-directed reasoning in the job-shop scheduling domain. The job-shop scheduling problem is defined as: selecting a sequence of operations whose execution results in the completion of an order, and assigning times (i.e., start and end times) and resources to each operation. The number of possible schedules grows exponentially with the number of orders, alternative production plans, substitutable resources, and possible times to assign resources and perform operations. The acceptability of a particular schedule depends not only on the availability of alternatives, but on other knowledge such as organizational goals, physical limitations of resources, causal restrictions amongst resources and operations, availability of resources, and preferences amongst alternatives. By viewing the scheduling problem from a constraint-directed search perspective, much of this knowledge can be viewed as constraints on the schedule generation and selection process. The problem of scheduling orders in a job-shop under these constraints raises a number of issues of interest to the artificial intelligence community such as: (1) knowledge representation semantics for organization modeling. (2) extending knowledge representation techniques to include the variety of constraints found in the scheduling domain, (3) integrating constraints into the search process--in particular, determining how to use constraints to bound the generation and focus the selection of alternative solutions, (4) relaxing constraints when conflict occurs, and (5) analyzing the interaction between constraints to diagnose poor solutions. In this thesis, we present a system called ISIS. ISIS uses a constraint-directed search paradigm to solve the scheduling problem. ISIS provides: (1) a knowledge representation language (SRL) for modeling organizations and their constraints, (2) hierarchical, constraint-directed scheduling of orders, which includes: (a) constraint-directed bounding of the solution space, (b) context-sensitive selection of constraints, and (c) weighted interpretation of constraints. (3) analytic and generative constraint relaxation, and (4) techniques for the diagnosis of poor schedules. In addition, the ISIS system has been designed to provide complete facilities for the practical use in the factory. These facilities include: interfaces for updating factory status, incremental scheduling in response to changes in the factory, interfaces for altering the factory model, and interactive, scheduling with flagging of poorly satisfied constraints. Versions of the ISIS program have been tested on a model of a real factory, using simulated orders.