变转速工况下基于改进奇异谱分解和1.5维包络阶次谱的风电机组轴承损伤识别

为实现变转速工况下风电机组轴承故障损伤的准确识别,提出一种基于改进奇异谱分解(ISSD)和1.5维包络阶次谱的诊断方法。针对奇异谱分解存在的端点失真和奇异谱分量数量判定问题,提出极限学习机延拓结合窗函数的端点效应抑制策略以及基于Person相关系数的分量数量判定策略。首先,通过计算阶次追踪算法对拾取的信号进行等角度重采样,继而对重采样角域信号进行ISSD处理;为便于后续分析,利用排列熵指标从ISSD处理结果中筛选出最佳主敏感奇异谱分量,对其执行对称差分能量算子解调运算,并计算所得包络信号的1.5维谱;最后通过分析1.5维包络阶次谱中的阶次成分准确判定轴承运行状态。实验台信号及实测工程信号验证表明,所提方法能有效提取变转速工况下风电机组轴承损伤特征,具有一定工程参考价值。

INJURY IDENTIFICATION OF WIND TURBINE BEARING BASED ON ISSD AND 1.5 DIMENSION ENVELOPE ORDER SPECTRUM UNDER VARIABLE ROTATING SPEED CONDITION

In order to achieve the accurate identification for the fault injury of wind turbine bearing under variable rotating speed condition,a diagnosis method based on improved singular spectrum decomposition and 1.5 dimension envelope order spectrum was proposed. Aiming at solving the problems of the endpoint distortion and the number determination of singular spectrum component,the endpoint effect suppression strategy combining the extreme learning machine extension and the window function was proposed. Firstly,the obtained signal was even-angle resampled using computed order tracking algorithm,then the resampled angle domain signal was processed by ISSD,and the component number determination strategy based on Pearson correlation coefficient was presented. In order to facilitate the subsequent analysis process,the permutation entropy index was used to select the optimal principal sensitive singular spectrum component from ISSD processing results,then the symmetrical differencing energy operator demodulation operation was carried out and the 1.5 dimension spectrum of the obtained envelope signal was calculated. Finally,the operating condition of bearing was accurately judged by analyzing the order components in the 1.5 dimension envelope order spectrum. The experimental signals and the measured engineering signals verification showed that,the proposed method can effectively extract the injury feature of wind turbine bearing under variable rotating speed condition and have a certain value for engineering reference.