机械振动数据的模型分析

目前，振动数据处理，包括谱分析，相关分析，系统分析等，广泛采用以FFT为基础的各种算法。本文讨论以建立数据模型为基础的数据处理，该方法有很多估点，文中还讨论了几个重要问题。关于模型阶次，本文指出，由信息论推出的判阶准则适合于短序列，可在非平稳振动信号分析中应用。对于平稳的振动信号，属于长序列，可使用更简单而有明确物理意义的准则。在不少情况中，对长序列模型阶次的判断并不是重要问题。笔者认为振动数据分析的研究重点是它的应用，并给出了应用实例。

MODEL ANALYSIS METHOD FOR MECHANICAL VIBRATION DATA PROCESSING

At present, the methods basing on FFT are employed wildly for vibration data processing, which includes the spectrum analysis, correlation analysis, test model analysis and parameter identification etc. The methods basing on data model for the data processing are presented in the paper. There are many advantages with the mode 1 analysis method. Some important criterions to judge the order of the model are appropriate to short time series, which can be come across in the unstationary random vibration data. The simpler criterions with more clear physical mean ought to be used for the stationary random vibration data. The application research ought to be the most important problem in the data model analysis. An application example is presented in the last part of this paper.