基于LMD模糊熵与超球结构支持向量机的水力机组涡带工况识别方法

Vortex Recognition of Hydropower Units Based on LMD Fuzzy Entropy and HSSVM

针对水力机组振动摆度信号的非平稳特性,提出了基于局域均值分解模糊熵及超球结构支持向量机的涡带工况识别方法。首先对振动摆度信号进行局域均值分解,获得一系列乘积函数分量;然后计算各乘积函数分量的相关系数,判断是否存在虚假变量;最后计算有效乘积函数分量的模糊熵作为特征向量,利用超球结构支持向量机进行小负荷区、涡带区和稳定区的机组运行工况识别。水电站水导轴承摆度信号试验结果表明,该方法能有效识别机组涡带工况区间。

Aiming at the non-stationary characteristics of vibration signal of hydropower units,a vortex recognition method was proposed based on the fuzzy entropy of local mean decomposition(LMD)and hyper sphere support vector machine(HSSVM).First,the vibration signal was decomposed by LMD,and a series of product functions(PFs)were extracted.Then,according to the correlation coefficient of each PFs and original signal,the false component was selected.Finally,the fuzzy entropies of principal PF components were calculated and set as the characteristic vector.The condition of small load area,vortex,and stable operation area of hydropower units can be recognized by the HSSVM.The experimental results show that the method can effectively diagnosis three kinds of working conditions.