【摘要】 研究非平稳加速度时间历程,有时并不需要知道随机变量的全部统计信息,而只需求得随机变量的某些既重要又有代表性的信息。基于兰帕尔-齐夫复杂度理论的符号基概念,由随机信号经粗粒化处理后得到符号序列,然后计算每个符号基的权系数,再将符号基线性叠加得到符号空间。将随机信号的采样数据分为n段再构建一个n行数据矩阵,计算此n行矩阵的符号空间。将该符号空间的系数矩阵进行奇异值分解,所得的奇异值即为频率,数据矩阵对应行的标准差即为相应频率的幅值。算例结果表明,该方法能够获得随机信号的代表性频率信息。对非平稳随机信号不需引入人为假定,可直接对数据进行计算。更多还原

【Abstract】 Some time it is not necessary to know all statistical information of the stochastic variable in studying of non-stationary acceleration time history,some representative information is simply to be known.Based on symbol group of Lempel-Ziv complexity,the symbol sequence and symbol group are been got from graining the random signal.The symbol space can be got by linearly plus all symbol groups.The sample data of random signal are been separated into n subsection,formation one n row data matrix.The singular value is been obtained by proceeding SVD of the coefficient matrix of the symbol space.The singular value is frequency;the stand and deviation of the row corresponding to the frequency in the data matrix is amplitude.A comparison of its results and that by FFT method shows that the proposed method is feasible.The data of non-stationary random signal can be analyzed at first hand,without to presume artificially.更多还原