一种基于高阶谱的平稳随机信号的高斯性检验算法

Gaussian Detection Algorithm Based on Stationary Signal of High Order Spectrum

为了提高短数据序列下对称分布和冲击型样本的高斯检验性能,提出了一种基于高阶谱的平稳随机信号的高斯性检验算法。采用高斯核函数进一步分析了高斯分布的三谱特性。在此基础上,以反映三谱波动性的方差为统计量建立了一个新的三谱检验模型,将该三谱检验模型与双谱秩和检验模型相结合组成一个高阶谱检验模型。仿真表明,该模型在短数据序列下可以取得比传统模型更好的高斯性检验结果。

A Gaussian detection algorithm based on stationary signals of the high order spectrum is proposed to improve the Gaussian detection performance on symmetrical distribtution and impact sample for a relative small data size. The tri-spectrum of Gaussian distribution is further analyzed by the Gaussian kernel estimator. On this basis, a new tri-spectrum detection model based on variance is established. Then, the tri-spectrum model is combined with the bispectrum rank model to form a high order spectrum detection model. Simulation results show that the high order spectrum model has a better Gaussian detection result than that of conventional models on a relative small data size.