基于集合经验模态分解和支持向量机的飞轮组件故障诊断

Fault Diagnosis of Flywheel Components Based on Ensemble Empirical Mode Decomposition and Support Vector Machine

卫星用飞轮组件是卫星姿态控制的重要执行部件,需要及时对飞轮组件运行状态进行监测,但实际生产过程中往往由于前期缺少试验数据无法对飞轮组件状态监测。因此本文提出使用加入白噪声的仿真数据优化支持向量机达到对飞轮组件进行状态监测。同时使用集合经验模态分解对实测信号降噪,使用降噪后的信号验证支持向量机的分类准确率。试验表明支持向量机能够识别正常状态、内圈故障状态和外圈故障状态,且分类有效率达到98.33%。

The flywheel assembly for satellite is an important executive component of satellite attitude control.It is necessary to monitor the status of flywheel components in time,but in the actual production process,it is often impossible to monitor the status of flywheel components due to the lack of experimental data in the early stage.It was proposed that the simulation data was used with white noise to optimize the support vector machine to achieve the purpose of state monitoring of flywheel components.At the same time,the ensemble empirical mode decomposition was used to denoise the measured signal.The denoised signal was used to verify the classification accuracy of the support vector machine.The test results showed that the support vector machine can identify normal,inner and outer ring fault states.The classification efficiency reached 98.33%.