Abstract
Fusion of heterogeneous data in real time is an important task in the diagnosis of technical objects. This is due to the need of taking into account not only the data coming from the sensors, but also external factors affecting the technical object. The article addresses problems of data fusion terminology on the basis of a review of the national and foreign literature. New definition of the term "data fusion" is proposed. Different scientific views of domestic and foreign experts on a problem of data fusion for diagnosing of technical objects in presence of different types of sensors and heterogeneous information are generalized and systematized. The adapted classification of data fusion, taking into account various criteria (the relations between sensors, the level of abstraction of data, architecture type), is presented. Classification of structural models of data fusion, developed for creation of intellectual systems of data fusion is given. Models of process of data fusion are investigated, their structures are presented, model merits and demerits are revealed. It is noted that for effective collecting of the basic data, arriving from of different types of sensors, and its processing it is possible to use several models of data fusion or their combination. It will allow making scientifically based management decisions when diagnosing difficult technical objects. While a lot of works of foreign researchers are devoted to separate sections of data fusion technologies, the paper presents the first full research in Russian where all the main aspects that belong to the intellectual technologies of data fusion are considered.