基于最小熵解卷积的风电机组故障诊断研究

发布时间：2018-06-06 20:10

本文选题：风电机组 + 故障诊断　；参考：《华北电力大学(北京)》2016年硕士论文

【摘要】：风力发电正在电力行业中占有越来越重要的位置。但是由于载荷复杂多变、运行环境条件恶劣,使得风电机组故障率以及运行维护费用较高。因此对风电机组进行状态监测与故障诊断具有重大的工程实际意义。然而在强噪声、强干扰、强耦合的背景下进行滚动轴承、齿轮等部件故障特征信息的早期提取一直是风电机组故障诊断领域的一大难题。最小熵解卷积方法正是一种十分便捷和有效的故障特征提取和智能诊断方法,通过最小熵解卷积方法来增强故障信号的冲击性,降低传递路径等因素带来的干扰,将更加有利于对设备故障早期特征信息的识别,从而提高故障诊断的准确性和及时性。本文主要研究基于最小熵解卷积的轴承故障特征增强方法及其在实测信号分析中的应用,主要研究内容：(1)对最小熵解卷积方法的理论基础进行探究,用仿真信号及风电机组实测信号验证了其有效性。(2)验证当对象信号的转速变化时,最小熵解卷积方法依然可以进行稳定可靠的状态判断。通过更改滤波器的参数,探究各个因素对最小熵解卷积结果的影响,并给出合理的参数设置。(3)对经过最小熵解卷积后信号的时域特征值与频域特征值的变化情况进行研究,并介绍了自组织特征映射(SOM)神经网络的基本理论,将用最小熵解卷积方法处理过的振动信号输入到SOM神经网络中进行训练,可以取得更理想的分类结果。
[Abstract]:Wind power generation is playing an increasingly important role in the power industry. However, due to the complex load and poor operating environment, the failure rate and maintenance cost of wind turbine are high. Therefore, the condition monitoring and fault diagnosis of wind turbine is of great practical significance. However, in the background of strong noise, strong interference and strong coupling, the early extraction of fault feature information of rolling bearings, gears and other components has been a major problem in the field of wind turbine fault diagnosis. The minimum entropy deconvolution method is a very convenient and effective method for fault feature extraction and intelligent diagnosis. The minimum entropy deconvolution method is used to enhance the impact of the fault signal and reduce the interference caused by the transmission path. It will be more helpful to identify the early feature information of equipment fault and improve the accuracy and timeliness of fault diagnosis. This paper mainly studies the bearing fault feature enhancement method based on minimum entropy deconvolution and its application in the analysis of measured signals. The main research content is: 1) the theoretical basis of the minimum entropy deconvolution method is explored. The validity of the method is verified by the simulation signal and the measured signal of the wind turbine. It is verified that the minimum entropy deconvolution method can still be used to judge the stable and reliable state when the speed of the object signal changes. By changing the parameters of the filter, the influence of each factor on the result of minimum entropy deconvolution is explored, and the reasonable parameter setting is given. The variation of the time domain characteristic value and the frequency domain characteristic value of the signal after the minimum entropy deconvolution is studied. The basic theory of self-organizing feature mapping (SOM) neural network is introduced. The vibration signal processed by minimum entropy deconvolution method is input into SOM neural network for training, and more ideal classification results can be obtained.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM315

【相似文献】