基于双相干谱的减压阀故障诊断

Fault Diagnosis of Reducing Valve Based on Bispectra Slices

双谱包含了信号的非对称、非线性信息,可以用来描述非线性相位耦合,尤其是二次相位耦合;双相干函数可以定量估计二次相位耦合的耦合程度;正常信号和故障信号的双相干谱呈现了不同的波峰特性;二维小波多级分解能在提取信号频率信息的基础上,有效对二维数组信息进行压缩;最小二乘支持向量机可以在有限样本下得到全局最优,从而避免局部最优问题,且具有降低计算复杂度的优点;实验中由获得的26组实验数据,利用二维小波多级分解对双相干谱进行特征提取,并输入LSSVM对减压阀进行故障诊断,取得了正确率接近90%的良好效果。

Bispectrum Includes nonsymmetric, nonlinear information of singals, it can be used to describe nonlinear phase couplings, especially quadratic phase couplings. Bicoherence function can evaluate the degree of quadratic phase couplings quantitatively, the hlcoherence spectrum of normal and fault singals show different peaks characteristics. 2-D wavelet multi-level decomposition can compress the informa- tion of 2--D arrays while extracting features of frequences in primitive signals effectively. LSSVM can find the global optimization solution with small samples, so avoids local optimization, and it can reduce computation complexity. 26 groups of data obtained from the experiment were used to extract features by 2-D wavelet multi level decomposition from bicoherence spectrum, then the features were input to the LSSVM, and diagnose the valve＇ s faults, the correct rate is nearly 90%, it is satisfactory.