【摘要】 本文详细分析了现有输入噪声模型和输出噪声模型的抗噪声原理,提出了一种能同时抑制输入噪声和输出噪声的无偏差估计模型,它利用二次互功率谱代替通常估计中的自功率谱。使得由于噪声引起的误差可以通过多次平均完全消除。新估计模型的精度与输入噪声和输出噪声的水平无关,仅与统计次数有关。在输入、输出噪声不相关条件下,即使相干函数很低仍能通过多次平均获得具有足够精度的频响函数值。文章最后给出了不同噪声干扰下的(?)估计值、(?)估计值以及本文模型的(?)估计值,并与没有噪声干扰下的H_o 值进行比较。更多还原

【Abstract】 This paper analyses the noise refrain principle of current Input and Output Noise Models in detail,andfind the reason which causes the deviation is due to the Auto-power Spectrum,in which the noise can’t beeliminated by averaging.On this base,the paper develops a new kind of noise refrain model,it uses the“twice Cross-spectrum”instead of Auto-power spectrum,so the deviation caused by the noise can be elimi-nated fully by averaging.The new model can refrain Input and Output Noise without considernation ofnoise levels,and the accuracy only has relation with averaging times.On condition that Input Noise isn’tcoherent with Output Noise,even the Coherence function is very low,the Frequency Response Functioncan also be estimated at good accuracy.At the end of the paper,the results estimated by the new model andthe existing ones are presented.更多还原