Data fusion for adaptive control in manufacturing: Impact on engineering information models

Abstract

Data fusion is the integration and analysis of data from multiple sensors to develop a more accurate understanding of a situation and determine how to respond to it. Although data fusion can be applied in many situations, this paper focuses on its application to manufacturing and how it changes some of the more traditional, less adaptive information models that support the design and manufacturing functions. The paper consists of four parts: Section 1 defines data fusion and explains its impact on manufacturing. Section 2 describes an information system architecture and explains the natural language-based information modeling methodology used by this research project. Section 3 identifies the major design and manufacturing functions, reviews the information models required to support them, and then shows how these models must be extended to support data fusion. Section 4 discusses the future directions of this work. This report is one of three produced by an FY93 LDRD project, Information Integration for Data Fusion. The project confirmed: (1) that the natural language-based information modeling methodology could be used effectively in data fusion areas, and (2) that commonalities could be found that would allow synergy across various data fusion areas, such as defense, manufacturing, and health care. The project found five common objects that are the basis for all of the data fusion areas examined: targets, behaviors, environments, signatures, and sensors. Many of these objects and the specific facts related to them were common across several models and could easily be reused. In some cases, even the terminology remained the same. This commonality is important with the growing use of multisensor data fusion. Data fusion is much more difficult if each type of sensor uses its own objects and models rather than building on a common set. Information model integration at the conceptual level is much easier than at the implementation level.