基于1(1/2)维倒谱的汽车齿轮变速器信号特征提取

Signal Feature Extraction for Vehicle Transmission Based on the Concept of 1.5 Dimensions Cepstrum

从三阶累积量出发,提出112维倒谱的概念,从理论上推导了其能抑制噪声、提高信噪比。并将其与倒谱对比,应用于实际的汽车齿轮变速器故障特征提取,分析表明,相对于倒谱而言112维倒谱能有效地从复杂的信号成分中提取出特征信号来,并且计算量小,精度高。

A concept of 1.5 dimension cepstrum based on third order cumulant is proposed. It is deduced from theory that 1.5 dimension cepstrum can suppress the noise and increase the single-to-noise ratio. Analysis indicates that using 1.5 dimension cepstrum can extract feature signals (transmission fault feature in this case) from the complex signal components with higher accuracy but much less computation effort, compare with the general cepstrum.