A New Data Mining Approach with Application in Satellite Fault Diagnosis

Abstract

For carrying out a real-time and effective fault diagnosis to satellite, this paper studies a new data mining approach that is called the fuzzy incomplete approach, and gives its algorithm implementation in the satellite fault diagnosis. The application comparison of the new data mining approach with other data mining approaches is discussed by measuring the performance of each approach from the error classified rate, missed classified rate, accuracy rate, and run-time. On the data of 1020 samples, the test results show that the error classified rate of the fuzzy incomplete approach is or so 4.12%, the missed classified rate is or so 1.18%, the accuracy is or so 94.71%, the run-time is 0.35s. The experiment results show that the performance of the fuzzy incomplete approach is better. These researches in this paper offer a suitable data mining approach for accurately distinguishing the fault characteristics of satellite.