柴油机振动信号谱估计的自回归方法

介绍了经典谱分析和现代谱分析中自回归 (AR)谱分析的各种常用方法 ,比较了各种方法对模拟信号 (样本总数N =32 )和柴油机爆发信号 (N =12 8)的处理效果。结果表明 ,经典谱分析方法的稳定性好 ,但分辨率受信号样本数影响 ,不适于小样本信号。现代谱分析方法比经典谱分析方法有更高的分辨率 ,因而可用于柴油机小样本数据段的分析 ,但使用时要注意模型阶次 p的选择。当 p太高时 ,各种方法都出现不稳定的现象。自相关法的分辨率略低 ,但其稳定性较好 ,可优先考虑使用。柴油机振动过程中有许多小样本信号 ,如爆发、阀开关、喷油信号等 ,它们引起的振动信号都非常短 ,采用AR谱估计方法是非常合适的

Autoregression spectra analysis of vibration signals of diesel engine

The common methods applied to autoregression spectra analysis were discussed. The results of different methods for treating analog signals (32 samples) and breaking signals (128 samples) of diesel engine were compared. The classical method is stable, but the number of the signal sample affects its resolution. So, the classical method is not suitable to the small-sample signals. The modern method has higher resolution than the classical method, and it is available to analyze the small-sample signals from the vibration of diesel engine. The power of the model should be carefully selected in the application of the modern method. When the power is too high, the unstable conditions may happen for various methods of autoregression spectra analysis. The autocorrelation method has the lower resolution power and priority in use because of its more stability. Many small-sample signals may be produced in vibration of diesel engine and have short vibration signal, for which the autoregression spectra estimation is suitable.