Temporal Constraint Propagation for Airlift Planning Analysis.

摘要

Developing efficient airlift plans for large operations is difficult even for experienced planners. Time is often critical and days or hours may make the difference between success and failure. Airlift plans are developed and refined through a repetitive cycle to produce usable schedules. A planner selects resources for a plan, develops a trial schedule, and analyzes the schedule for weaknesses. This process is very time-consuming and a method is needed to analyze airlift plans and provide useful feedback early in the planning process. Temporal reasoning provides a general mechanism for such analysis. Different types of temporal constraints can be inserted into a network of airlift events to provide time bounds on execution of the complete plan. For this purpose we developed a general temporal constraint reasoner and a set of mechanisms for deriving temporal information from airlift requirements and partial schedule specifications. Physical limitations of the aircraft and operating facilities as well as the availability of cargo all provide constraints on when certain events may occur. These constraints may be the time required to fly from one location to another or the time spent waiting for an aircraft to be loaded. Comparing cargo requirements with airlift capacity over time provides additional constraints. The advantage of using a temporal constraint network as the underlying representation is its ability to accommodate various sources of information about time relationships between events in a plan. By asserting temporal information about specific events in an airlift plan, the planner can assess the impact of high-level planning decisions. Theses.