基于倒双谱的机械系统载荷识别方法

提出了一种基于倒双谱的机械系统载荷识别的新方法,首先,确定系统输出序列及系统特性的表示方法,通过系统输出序列的倒双谱及系统特性的微分倒谱,建立系统输出与系统特性的参数联系,然后,通过参数估计辨识得到系统特性,再通过时序分析的方法,由已知的系统特性求得其“逆传系统”特性,并利用此“逆传系统”特性和该系统输出序列来确定系统载荷,该方法解决了在系统特性和系统载荷未知的情况下,能量有限的稳定单输入－单输出因果线性系统的载荷识别问题,最后,以随机载荷作用下的单输入－单输出系统为例,采用最小平方算法,通过系统输出序列的高阶累积量直接估计系统特性的微分倒谱参数,辨识得到系统特性,再利用“逆传系统法”进行载荷识别,仿真验证了该理论方法的有效性。

Dynamic Load Identification Method Based on Bicepstrum

A new method based on bicepstrum was proposed for dynamic load identification. Firstly, the parameters relationship between the output sequence and the impulse response of a system was established through bicepstrum of the higher order cumulants of the output sequence and the differential cepstrum of the system properties. Then the system property can be identified by estimating the parameters. The parameters of an "inversion system" of the known system characteristic can be obtained with time serials analysis method. Finally, dynamic load was determined by using the output sequences and the inversion system. It is a practicable solution for load identification of any stable causal linear single-input/output system with limited energy without a priori knowledge of the characteristic and load of the system. An example of single-input/output system under random load was analyzed by computer simulation. With higher order cumulants of the output sequence, differential cepstrum of the system properties was estimated directly based on the method of least-squares (LS) arithmetic. Then the system characteristic was identified. Inversion system method was used to determine the load. This example demonstrates the effectiveness of the new dynamic load identification method.