A generalized rough set-based information filling technique for failure analysis of thruster experimental data

Interval-valued data and incomplete data are two key problems for failure analysis of thruster experimental data and have been basically solved by the proposed methods in this paper. Firstly, information data acquired from the simulation and evaluation system formed as interval-valued information system (IIS) is classified by the interval similarity relation. Then, as an improvement of the classical rough set, a new kind of generalized information entropy called “H′-information entropy” is suggested for the measurement of uncertainty and the classification ability of IIS. There is an innovative information filling technique using the properties of H′-information entropy to replace missing data by some smaller estimation intervals. Finally, an improved method of failure analysis synthesized by the above achievements is presented to classify the thruster experimental data, complete the information, and extract the failure rules. The feasibility and advantage of this method is testified by an actual application of failure analysis, whose performance is evaluated by the quantification of E-condition entropy.