机械振源数估计的四线性平行因子方法研究

Quadrilinear parallel factor method for estimating the number of mechanical vibration sources

  • 摘要: 针对三线性平行因子模型在机械振源数盲估计方法中存在的不足,即由于维度的限制只能对振动信号的部分信息建模,信号中的时间序列信息被忽略。将三线性平行因子的机械振源数估计方法扩展至四维,提出一种基于四线性平行因子的机械振源数估计方法。所提出的方法中,在三维的基础上增加了第四个维度(时间维),利用四线性交替最小二乘法迭代更新载荷矩阵,用核一致诊断法估计振源数。提出的方法既继承了三线性平行因子的独特优势,同时,又包含了更完整的振动信号的信息。相比三线性平行因子机械振源数估计方法,同组分数条件下四线性平行因子比三线性平行因子得到的核一致值更高,估计更加准确。仿真结果表明,提出的方法优于三线性平行因子机械振源数估计方法。实验结果进一步验证了提出的方法的有效性。

     

    Abstract: Based on the deficiencies of the three-linear parallel factor model in the blind estimation method of the number of me-chanical vibration sources,that is,due to thelimitation of the dimension,only part of the information of the vibration signal can bemodeled,and the time series information in the signal is ignored.In this paper,the estimation method of mechanical vibrationsources number based on the trilinear parallel factor is extended to four dimensions,an estimation method of mechanical vibrationsources number based on the quadrlinear parallel factor is proposed.In the proposed method,the fourth dimension(i.e.time dimension)is added on the basis of three-dimensional,the load matrix is iteratively updated by the four linear alternating leastsquares method,and the number of vibration sources is estimated by the kemel uniform diagnosis method.Compared with the estimation method of mechanical vibration sources number based on the trilinear parallel factor,the proposed method inherits theunique advantages of the trilinear paralel factor,and contains more complete information of the vibration signal.Moreover,underthe condition of the same component number,the quadrlinear parallel factor has higher coincidence value than the trilinear parallelfactor,and has the more accurate estimation.The simulation results show that the proposed method is superior to the estimationmethod of mechanical vibration sources number based on the trilinear paralel factor.The experimental results further verify the effectiveness of the proposed method.

     

/

返回文章
返回