A Kind of Faults Knowledge Discovery Pattern by Means of Rough Set Theory

Abstract

The knowledge acquisition in expert system used to faults identification of mechanical device is very difficult. The puzzle has become the critical obstacle in the developing progress of machinery information technology. So based on the classification concepts of Rough Set Theory (RST) and the thinking of big data, the way to develop the technology along the data driven way was explored in this paper. By the concepts of classification contained in RST, the data operation scheme on knowledge conversions was present. It shows that there are two kinds of relations in the process of knowledge conversions, i.e., the knowledge equivalent relation and the knowledge inclusion relation. They indicate that to realize the knowledge acquisition is a systematized engineering project according to big data idea and by data-driven methods. Yet the primary task is to build a suitable data structure model and to use it accumulate the original knowledge of faults diagnosis with the specialized data mode. After obtaining the huge amounts of data contained the original decision knowledge, the other jobs includes that looking for all kinds of algorithms reduces the data size and achieves eventually the knowledge discovery. To solve well the faults knowledge acquisition in by the data driven of big data technology, the conclusion is that to establish a kind of valuable data structure model store the original faults decision knowledge from factory site is a most critical procedure. To carry out the data classification and clustering analysis using some intelligent tools like RST, it is a basic requirement to build a special data structure model for specific domain objects.