Research On Space Bearing Life Prediction Method Based On Optimized Support Vector Machine

The long life and high reliable research of the space bearing is one of the basis forthe development of long-life spacecraft. The life assessment and prediction research ofthe space bearing is the most importment content of the space bearing long life and highreliable research. The space bearing needs to bear alternating temperature, high-energyparticle irradiation, the combined effects of the atomic oxygen erosion, debris impactand dust erosion and other extreme environments, the failure behavior and mechanismis very different from the bearing working in the conventional terrestrial environment,however, there lack of depth understanding of the laws. Currently, we just only throughthe accelerated life testing group in the simulated space environment to get the failuredata, in order to do the space bearing life assessment and prediction research. The paperchosen the vibration signal which indulgences the rich life evolution information towork as the object of the study, due to the friction torque can not effectively reflect thespace bearing life evolution characteristics. By researching the failure mechanism thatleading to the space bearing performance degradation accuracy, this research reveals thevibration characteristics can reflect the space bearing degradation trends. By processingthe space bearing vibration signal, this research extracts the index to reflect the spacebearing degradation process. By building the artificial intelligence model, this researchachieves the space bearing life prediction. All of these researchs will have greatsignificance for guiding the design and manufacture of space bearing, establishing thelife assessment criterion, improve the spacecraft performance subsequently. Thisresearch contents are as follows:①Because the failure mechanism of the space bearing is very complex under theextreme environment, and the failure characteristics is difficult to reflect the bearingdegradation process, this article researchs the failure mechanism that leading to thespace bearing performance degradation accuracy, establishs the line between thevibration characteristics and the space bearing degradation trends. Reveals that thevibration characteristics changed with the bearing degradation process, this will laid thetheoretical foundation for using the vibration signal to predict the space bearing life.②Aiming at the problem that the space vibration signal collected from thesimulated space environment experiments contains serious background noise. Thisresearch uses the blind source separation method to eliminate the background noise. Through researchs the components and characteristics of the background noise,constructs the virtual signals and builds the blind source separation model to eliminatethe noise. The application results showed that the method can eliminate the backgroundnoise effectively.③Aiming at the problem that the space bearing degradation process indicator isvery difficult to been built. This paper constructs the indicator based on the principalcomponent analysis weighted fusion method. Researchs the time domain, frequencydomain and time-frequency domain feature information extraction methods, exploresthe spatial distribution of these features. Constructs the indicator based on the dataspace mapping and weight fusion method, the results show that the constructedindicator can reflect the bearing degradation trend effectively.④Aiming at the traditional life prediction method can not predict the spacebearing life effectively, this article uses the optimizatic SVM to predict the spacebearing life. Uses the phase space reconstruction method to select the inpur parametersof the SVM, uses the particle swaram algorithm to select the SVM internal parameters.Through the optimized SVM model to achieve the space bearing life prediction, resultsproved the effectiveness of the proposed method.⑤This research make the space rolling bearing performance degradation trendforecasting soft module and space rolling bearing residual life prediction soft module,human-computer interaction module, so as to meet the need of space bearing lifeprediction function.Finally, this work is summarized.