风电机组轴承早期故障特征提取研究

针对风电机组滚动轴承早期故障振动信号微弱、强干扰、非平稳、非线性的特点,提出基于自适应噪声完整集成经验模态分解(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise,CEEMDAN)-排列熵(Permutation Entropy,PE)-遗传算法(Genetic Algorithm, GA)的特征提取方法。方法先计算振动信号经CEEMDAN分解得到多个本征模态函数(Intrinsic Mode Function,IMF)的排列熵值和方差贡献率,剔除虚假、低贡献率分量;根据识别误差最小和特征子集数目最少两个目标,构造了适应度函数,通过GA进行特征选择选出最优特征子集。仿真分析,上述方法能够快速有效提取不同故障的振动信号特征指标,为故障模式识别问題提供良好的思路和方法。

Research on Early Fault Feature Extraction of Wind Turbine Bearing

Aiming at the characteristics of weak,strong interference,non-stationary and non-linear vibration signals of wind turbine rolling bearings in the early stage of fault,a new method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)-Permutation Entropy(PE)-Genetic Algorithm(Ge)is proposed.The feature extraction method of netic Algorithm,GA.Firstly,the vibration signals are decomposed by CEEMDAN to get the ranking entropy and variance contribution rate of multiple intrinsic mode functions,and the false and low contribution rates are eliminated.Select the best characteristic subset.Simulation results show that this method can quickly and effectively extract the vibration signal characteristics of different faults,and provide a good idea and method for fault pattern recognition.