TACTICS: A knowledge-based tactical production planning system.

摘要

Tactical Production Planning (TPP) is a process that involves transforming market information in the form of demand forecasts into capacity requirements to produce a smoothed production plan in light of multifunctional and top management objectives. Tactical production planning has been traditionally treated in the Management Science literature under the name of Aggregate Production Planning (APP). In general, the APP models has a narrow view of the process involved and has had limited applications due to their simplifying assumptions. Current trends in manufacturing also render these models obsolete.; This research identifies the characteristics of tactical production planning in a modern manufacturing firm and proposes a new conceptual model. The second phase of this dissertation develops a knowledge based system (TACTICS) which incorporates the actual heuristics and strategies used by planners of Engineering & Manufacturing Columbia (a division of NCR Corporation) to prepare tactical production plans.; It was found that tactical production planning in a modern manufacturing firm involves a process which is substantially different from the traditional formulation of the problem: the characteristics of the TPP process are deeply rooted in the way this company goes about doing business and the industry's environment it faces.; TACTICS models and solves an organizational problem integrating the decisions made by two human experts. These two experts work on different parts of the TPP problem and belong to different departments. TACTICS is organized as a partitioned rule base and utilizes meta-rules and a hierarchy of sponsors to control the reasoning process. Frames represent the descriptive knowledge and a hierarchy of terms extracted from the verbal protocol analysis represents data fussiness or vagueness in the system.; TACTICS performs very close to the experts from which the knowledge was acquired and is potentially more consistent and systematic. Furthermore, this system goes beyond the simple emulation of experts to support the manager decision process by providing alternative solutions based on the humans' reasoning logic.