基于小波包分析及神经网络的汽轮机转子振动故障诊断  

Turbine rotor vibration faults diagnosis based on wavelet packet analysis and neural network

作　　者：梁平[1] 白蕾[1] 龙新峰[2] 范立莉[3] 

机构地区：[1]华南理工大学电力学院,广东广州510640 [2]华南理工大学化工与能源学院,广东广州510640 [3]广东电网公司电力科学研究院,广东广州510640

出　　处：《控制理论与应用》2007年第6期981-985,共5页Control Theory & Applications

摘　　要：根据Bently实验台所采集的碰摩、松动、不对中、不平衡4种典型汽轮机转子振动故障信号,运用小波包分析方法对其进行能量分析并提取故障特征.分析结果表明:小波包分析与信号能量分解的故障特征提取方法,可以获得汽轮机转子振动的故障状态,有较好的故障区分度;另外由于经过小波包分解再重构后所提取的故障特征参数浓缩了汽轮机转子振动故障的全部信息,而BP神经网络具有优良的非线性映射能力,对提取的故障特征参数应用BP神经网络映射,可对汽轮机转子振动故障进行进一步的诊断.诊断结果表明:基于小波包分析及神经网络的故障诊断方法,具有较高的故障识别能力.According to the four typical fault signals of turbine vibration including rubbing, loosing, misalignment and mass unbalance collected from the Benfly experiment table, energy analysis and symptom extraction are carried out by wavelet packet analysis. The results of analysis indicate that symptom extraction by wavelet packet analysis and energy decomposition can obtain the faults state of turbine rotor vibration, possess better differentiation capability of fault types. In addition, the fault symptom parameters extracted by wavelet packet decomposition and reconstruction condense the whole information of turbine rotor vibration faults, and neural network possesses good non-linear mapping capability. For these symptom parameters, applying BP neural network mapping can diagnose the turbine rotor vibration faults further. The results of diagnosis indicate that the faults diagnosis method based on wavelet packet analysis and neural network has better faults identification capability.

关 键 词：小波包分析 汽轮机转子 故障诊断 特征提取 BP神经网络 

分 类 号：TK268[动力工程及工程热物理—动力机械及工程]