风力发电机组故障特征及基于警报信号的故障诊断研究

发布时间：2018-01-21 22:23

本文关键词： 风力发电机组不同控制策略故障表征 SCADA警报信号 DS证据理论故障诊断可靠性　出处：《南京理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着风电发展规模的不断扩大和技术发展的日趋成熟,提高风电设备的可靠性将成为我国风电领域发展要求和必然趋势。本文主要对风力发电机组的不同控制策略下的故障表征提取和基于警报信号的故障诊断进行研究,致力于提高风电机组的可靠性。为了对比不同控制策略下,风力发电机组的故障表征,本文利用MATLABSimulink建立分别建立最优尖速比控制策略和最优转矩控制策略下的风力发电机组整机模型,以齿轮箱故障引起的传动效率降低为故障模式,分别提取健康状态下和齿轮箱传动效率90%和80%的功率、电机转矩和机械转矩,然后,对不同控制策略下的这些故障表征做对比,并分析结果的成因,为后面的警报信号的故障诊断研究做准备。警报信号作为SCADA数据的一部分常常的得不到有效利用,本文对比了传统的高频信号故障诊断方法,提出了基于低频信号SCADA警报信号的故障诊断方法。警报并不等同于故障,由于风电系统是一个复杂的系统且系统内各个子部件联系紧密,当发生故障时,其之间相互影响,进而导致批量警报的发生。因此,本文通过模糊决策方法研究出一种根源警报查找方法,可以在大量警报中找出根源警报所在。然而,这时警报并没有和故障联系,无法使维修者提前制定维修方案,因而本文又提出了一种DS证据理论的故障诊断方法,首先将维修记录中的故障作为样本空间,利用维修记录和警报信号利用贝叶斯方法提取基本概率分配函数(BPA),然后利用改进后的DS证据理论进行故障诊断,使警报与故障联系,维修者可以从警报信号中诊断出故障的具体形式,从而指导维修者及时制定维修方案,节约时间和成本。最后,本文将DS证据理论故障诊断方法,进行MATLAB GUI界面化编程,并对该方法进行了横向验证,验证了该方法具有可行性、有效性和准确性。通过本文的研究,为未来风力发电机组的可靠性提高提供了充足的理论基础。
[Abstract]:With the continuous expansion of the scale of wind power development and technology development is becoming more and more mature. Improving the reliability of wind power equipment will become the development requirement and inevitable trend of wind power field in China. This paper mainly studies the fault representation extraction and fault diagnosis based on alarm signal under different control strategies of wind turbine. Research. To improve the reliability of wind turbine units, in order to compare the different control strategies, wind turbine fault characterization. In this paper, MATLABSimulink is used to establish the wind turbine model under the optimal speed ratio control strategy and the optimal torque control strategy, respectively. The transmission efficiency caused by the gearbox failure is reduced as the fault mode, and the power, motor torque and mechanical torque of the gearbox transmission efficiency 90% and 80% in the healthy state and the gearbox transmission efficiency are extracted respectively. Then the motor torque and the mechanical torque are extracted. The fault characteristics under different control strategies are compared, and the causes of the results are analyzed. As part of the SCADA data, the alarm signal is often not used effectively. This paper compares the traditional high-frequency signal fault diagnosis method. A fault diagnosis method based on low-frequency signal SCADA alarm signal is proposed. Warning is not equivalent to fault because wind power system is a complex system and each sub-component of the system is closely related. When the fault occurs, it affects each other, which leads to the occurrence of batch alerts. Therefore, this paper studies a method of finding source alerts by fuzzy decision method. The source alert can be found in a large number of alerts. However, at this point, the alarm is not linked to the failure and does not enable the repairman to develop a maintenance plan in advance. Therefore, a fault diagnosis method based on DS evidence theory is proposed in this paper. Firstly, the fault in maintenance record is taken as sample space. The basic probability distribution function (BPAA) is extracted by using the Bayesian method and the improved DS evidence theory is used for fault diagnosis to link the alarm with the fault. The maintainer can diagnose the specific form of the fault from the alarm signal, so as to guide the maintainer to make the maintenance plan in time, save time and cost. Finally, the paper presents the fault diagnosis method of DS evidence theory. The interface programming of MATLAB GUI is carried out, and the method is verified horizontally, which proves that the method is feasible, effective and accurate. It provides sufficient theoretical basis for improving the reliability of wind turbine in the future.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM315

【相似文献】