Abstract:

A fault diagnostic approach is proposed based on consistent approximate indication space, to deal with incomplete and inconsistent information. Firstly, an inconsistent diagnostics information system is introduced to express the inconsistent information. And the inconsistent diagnostics information system is translated into consistent approximate indication space. Then attribute reduction is carried out. Secondly, a diagnostic method based on inclusion degree is designed for complete information objects. And vectors are complemented for incomplete-information objects. Finally, this approach is validated with aero-engine data. Results show that the designed approach can deal with incomplete and inconsistent information effectively, and enhance the accuracy of diagnosis remarkably.