Research on prognostics for degradation trend of gyroscope based on EMD-SOM quantization error model

Abstract

According to the characteristics of the performance parameters for gyroscopes' degradation, we put forward a degradation trend prediction method with combination of empirical mode decomposition(EMD) and self-organizing mapping (SOM) quantization error model (MQE)for them. Firstly, we collect some weak signals as feature information for gyroscopes' degradation states, and use EMD to realize adaptive feature extraction for eliminating some noise interference. Secondly, according to the different state features of gyroscope, we use SOM method to realize multi-sensor information fusion, while modeling quantization error to realize failure prediction of the gyroscopes. Lastly, the experimental results and verifications show that the EMD method fusing sensor signals can extract the feature information for gyroscopes' degradation states simply and effectively. The self-organizing mapping quantization error model can predict gyroscopes' degradation trend for. So the method can provide the basis for the gyroscopes' maintenance decision.