电力电子电路故障诊断与故障预测方法研究

发布时间：2018-04-15 03:28

  本文选题：电力电子电路 + 故障诊断　； 参考：《湖南大学》2016年硕士论文

【摘要】：随着电力电子技术的迅猛发展,电力电子设备越来越复杂,规模也越来越大,其发生故障的概率也越来越高。电力电子电路作为该设备结构中的重要部分,它的故障可能会导致整个系统的故障甚至是瘫痪,造成严重的损失。因此保证系统运行的可靠性与稳定性,及时地发现故障与预防故障的发生就显得尤为重要,也使得对电力电子电路进行故障诊断与预测方法的研究得到了越来越多的重视。本文针对故障诊断步骤中的故障特征提取与故障辨识,以及电路的故障预测方法进行了研究,包括以下内容：采用主成分分析方法对电力电子电路的故障特征进行提取,得到含有原始故障数据大部分信息的主成分,并将它们组合成为新的特征向量,既降低了数据维数也使特征得到了突显。然后研究了基于Fisher判别分析法的(Fisher Discriminant Analysis,FDA)故障辨识方法,利用它对故障特征进行识别,得到最后的故障诊断结果,还和RBF神经网络进行辨识的实验结果相比较。仿真实例验证了所提方法的有效性,诊断的准确率高。考虑到电力电子电路的非线性性质,以及容差、环境等因素的影响,特征数据间关系复杂。为了提高故障特征的辨识度,采用高阶累积量方法(High Order Cumulant, HOC)提取电路的故障特征,得到各样本对应的的峭度和偏度,并将它们组合成新的故障特征向量。再将处理得到的新的特征向量输入FDA故障辨识方法中进行识别,并与第二章诊断方法的准确率进行比较,仿真实例验证了所提方法的有效性并且具有较高的诊断准确率。为了预防电路故障的发生和及时地采取“预知”维修,对电力电子电路故障预测的方法进行了研究。首先使用主成分分析与HOC对电路各种状态下输出电压信号的特征数据进行提取,再通过FDA进行处理,将处理得到的数据构造为一个可以反映电路健康状态的故障指示参数。然后根据故障指示参数所显示出的退化趋势得到一个经验模型,接着研究使用粒子滤波方法进行故障预测和剩余有用寿命的估计。该方法实施简单,实例证明了所提方法的有效性。
[Abstract]:With the rapid development of power electronics technology, power electronic equipment is becoming more and more complex, the scale is becoming larger and larger, and the probability of failure is becoming higher and higher.Power electronic circuit as an important part of the equipment structure, its failure may lead to the whole system failure or even paralysis, resulting in serious losses.Therefore, it is very important to ensure the reliability and stability of the system, to find and prevent the fault in time, and to make more and more attention to the fault diagnosis and prediction methods of power electronic circuits.In this paper, fault feature extraction and fault identification in fault diagnosis step, and fault prediction method of circuit are studied, including the following contents: the main component analysis method is used to extract fault feature of power electronic circuit.The principal components containing most of the information of the original fault data are obtained and combined into a new feature vector which not only reduces the dimension of the data but also highlights the features.Then, the fault identification method based on Fisher discriminant analysis is studied. The fault features are identified by using the method, and the final fault diagnosis results are obtained, and the results are compared with the experimental results of RBF neural network identification.The simulation results show that the proposed method is effective and the diagnostic accuracy is high.Considering the nonlinear property of power electronic circuit and the influence of tolerance and environment, the relationship between characteristic data is complex.In order to improve the identification degree of fault features, the high order cumulant method is used to extract the fault features of the circuit, and the kurtosis and skewness of each sample are obtained, and they are combined into new fault feature vectors.Then the new eigenvector is input into the FDA fault identification method to identify, and compared with the accuracy of the second chapter of the diagnosis method. The simulation results show that the proposed method is effective and has a high diagnostic accuracy.In order to prevent the occurrence of circuit faults and adopt "predictive" maintenance in time, the method of power electronic circuit fault prediction is studied.Principal component analysis (PCA) and HOC are used to extract the characteristic data of the output voltage signal in various states of the circuit, and then processed by FDA, the processed data is constructed as a fault indicator parameter which can reflect the healthy state of the circuit.Then an empirical model is obtained according to the degradation trend shown by the fault indication parameters, and then the particle filter method is used to predict the fault and estimate the remaining useful life.The method is simple to implement, and the effectiveness of the proposed method is proved by an example.
【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM507
，

本文编号：1752331