基于证据理论的风电机组故障诊断研究

发布时间：2018-10-09 10:04

【摘要】：风电机组的故障诊断是风力发电中的一项重要技术。随着风电产业的蓬勃发展,风电机组故障诊断方面不断有新的、更高的要求被提出来。风电机组的故障诊断具有重大意义。证据理论在处理不确定性信息时更加方便快捷,广泛应用于根据多源信息进行故障诊断。风电机组故障多源信息表示、不确定性等特点使得证据理论成为融合诊断的一种合适选择。基于证据理论的多源信息融合方法,本文将该方法应用于风力发电机组故障诊断,具体包括以下三点:(1)介绍了信息融合技术基本原理与信息融合处理模型,并给出了信息融合在数据级、特征级以及决策级三个层次上的融合模型,分析了该技术时如何处理故障信息,并在故障诊断得到应用的。通过对证据理论基本框架和相应组合规则的研究,总结了相关的基于证据理论的故障诊断方法。针对风力发电机组故障诊断,分析了基于证据理论的多源信息融合方法的可行性。(2)通过对相对熵原理的研究应用,解决了证据理论融合过程中原始证据高度冲突对融合结果的影响。根据多个传感器得到的证据重要性不同,通过相对熵原理判断每个证据影响因子的大小,即通过赋予每个证据不同的权值,从而削弱或放大每个证据的影响力。将该方法应用于风电机组轴承故障诊断中,取得良好的诊断效果。该方法通过加速度传感器提取轴承故障振动信号,采用经验模态分解法(EMD)分解故障振动信号,采集包络谱特征频率,作为故障特征量。通过与各故障样本频率比较,做出初步诊断;引入灰色理论中的相关性原理,获取原始证据;利用相对熵原理对原始证据分配相应权值,最后通过D-S组合规则对修改后的证据进行融合。(3)在现实故障诊断中,外界环境的影响使得证据理论的应用存在一定的不足,即从不完整、不确定信息中提取BPA(基本概率赋值函数)还存在一定的限制。本文通过对基于随机集的信任测度和基于随机集的似真测度进行含糊化处理,该方法能够很好的提取证据理论的BPA(基本概率赋值函数);在对风力发电机进行故障诊断时也能够取得较好的诊断结果。通过比较,含糊化处理不确定性信息比模糊化处理不确定性信息的效果更好,更能客观全面地体现不确定信息的本质内容,大大提高故障的识别诊断能力。该方法不仅适用于风电系统的故障诊断,也可用于诊断其他的机电设备的不确定性故障。
[Abstract]:Fault diagnosis of wind turbine is an important technology in wind power generation. With the rapid development of wind power industry, new and higher requirements have been put forward for wind turbine fault diagnosis. The fault diagnosis of wind turbine is of great significance. Evidence theory is more convenient to deal with uncertain information and widely used in fault diagnosis based on multi-source information. The characteristics of multi-source information representation and uncertainty of wind turbine fault make evidence theory a suitable choice for fusion diagnosis. Based on the evidence theory, this paper applies this method to wind turbine fault diagnosis, including the following three points: (1) the basic principle of information fusion technology and the information fusion processing model are introduced. The fusion model of information fusion at data level, feature level and decision level is given, and how to deal with fault information in this technology is analyzed and applied in fault diagnosis. Based on the study of the basic framework of evidence theory and the corresponding combination rules, the relevant fault diagnosis methods based on evidence theory are summarized. For wind turbine fault diagnosis, the feasibility of multi-source information fusion method based on evidence theory is analyzed. (2) through the research and application of relative entropy principle, The influence of the high conflict of the original evidence on the fusion result is solved. According to the different importance of the evidence obtained by multiple sensors, the magnitude of each evidence influence factor is judged by the relative entropy principle, that is, by giving each evidence a different weight, the influence of each evidence is weakened or amplified. The method is applied to the bearing fault diagnosis of wind turbine, and good diagnosis effect is obtained. In this method, the vibration signal of bearing fault is extracted by acceleration sensor, and the fault vibration signal is decomposed by empirical mode decomposition (EMD) method, and the envelope spectrum characteristic frequency is collected as the fault characteristic quantity. By comparing with the frequency of each fault sample, we make a preliminary diagnosis; introduce the correlation principle of grey theory to obtain the original evidence; use the relative entropy principle to assign the corresponding weight to the original evidence. Finally, the modified evidence is fused by D-S combination rule. (3) in the actual fault diagnosis, the influence of external environment makes the application of evidence theory have some shortcomings, that is, never complete. There are some limitations in extracting BPA (basic probability assignment function) from uncertain information. This paper deals with the ambiguity of trust measures based on random sets and quasi-true measures based on random sets. This method can extract evidence theory of BPA (basic probability assignment function) and obtain good diagnosis results in fault diagnosis of wind turbine. By comparison, the effect of ambiguity processing uncertainty information is better than that of fuzzy processing uncertainty information, the essence of uncertain information can be reflected objectively and comprehensively, and the ability of fault identification and diagnosis is greatly improved. This method is not only suitable for fault diagnosis of wind power system, but also can be used to diagnose uncertain faults of other electromechanical equipment.
【学位授予单位】：上海电机学院
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM315

【相似文献】