摘要
为解决经典谱估计算法在汽轮机振动信号采样点数较少时应用存在局限的问题,文中采用了最小方差谱估计算法,通过信息论准则确定其最优阶次和加最优窗降低频率偏移和谱线分裂程度.仿真结果表明:最优阶次下的最小方差谱估计算法,其功率谱图要比经典谱估计算法与AR模型算法的功率谱图更加平滑;谱峰尖锐与AR模型的谱图尖锐程度基本一致.
The classical spectrum estimation algorithm has limits when there are fewer sampling points of the turbine vibration signal. To solve the problem, the paper adopts the minimum variance spectral estimation. Through determining its optimal order times by the Akaike Information Criterion(AIC) and adding the best window to reduce the frequency offset and line splitting degree,its variance performance is improved. The simulation results show that the variance performanc of Minimum-Variance Spectral Estimation(MVSE) algorithm with Optimal Order is better than those of the classical algorithm and AR model algorithm. The variance performance of minimun vartance spectral estionati Minimum-Variance Spectral Estimation(MVSE) algorithm is almost the same as that of AR model algorithm. Its sharp degree of power spectrum^s peak is almost the same as that of AR model algorithm.
出处
《西安工业大学学报》
CAS
2015年第11期883-887,共5页
Journal of Xi’an Technological University
基金
陕西省科技厅自然基金(2012JQ8008
2013JQ8048)
陕西省教育厅科技专项(2012JK0545)
关键词
汽轮机
振动信号
谱估计
最小方差
信息论准则
turbine
vibration signal
spectrum estimation
minimum-variance
akaike information criterion