水轮发电机组模型参数辨识与故障诊断方法研究

发布时间：2018-08-05 10:33

【摘要】：随着我国水电事业的不断发展,大型水电站陆续建成投运,水轮发电机组逐渐向着大容量、高水头方向发展,确保水轮发电机组的安全稳定运行对提升电站经济效益、保证电厂及流域安全有着重要的意义。水轮发电机组是一个受水力、机械、电气等多场耦合影响的强非线性复杂巨系统,其运行机理尚不完全明确,机组的建模及故障诊断一直是相关研究与工程应用的难点问题。在此背景下,深入研究水轮发电机组系统辨识理论,开展水轮发电机组建模和模型参数辨识研究,不仅对提高水轮发电机组控制品质、提升电能质量、维护电力系统稳定都有着重要的理论意义和工程应用价值,而且还可为水轮发电机组的故障诊断提供理论依据和技术支撑。本文围绕水轮发电机组模型参数辨识、机组精细化建模以及故障诊断中存在的问题,以建立精细化水轮发电机组模型和探索先进的故障诊断方法为目的,通过深入分析水电机组的实际运行特性,结合系统辨识理论、智能优化方法和动态系统辨识方法,建立了适用于机组控制和电力系统分析的水轮发电机组精细化模型；深入开展了基于模型的水轮发电机组故障诊断方法研究,引入了Volterra级数、广义频率响应函数(GFRF)和非线性输出频率响应函数等建模理论,结合参数辨识方法,建立了水轮发电机组模型,并通过对模型参数变化的分析,提出了有效的故障特征提取方法和故障识别方法。论文的主要研究内容及创新性成果如下： (1)深入研究了现有水轮机模型,总结归纳各种水轮机模型的适用范围及优缺点,针对导水机构水力损失不可忽略且模型机理复杂的问题,引入曲线拟合的方法,建立了考虑导水机构水力损失的水轮机精细化模型。同时,将人工鱼群算法引入到水轮机精细化模型参数辨识中,为改善该算法后期收敛速度慢、容易陷入“局部最优”等问题,提出一种改进的人工鱼群算法(IAFSA),旨在提高算法的收敛性与精度。所提方法实现了水轮机精细化模型与导水机构水力损失曲线同步辨识。 (2)为简化电力系统网络结构,满足电力系统分析的需求,对中小水电机组进行了等值建模研究,建立了更适合表征水轮发电机凸极效应的五阶发电机组等值模型,提出了基于研究系统内部多点PMU测量数据与联络线数据的中小水电机群等值模型辨识目标函数,同时改进了辨识策略以提高辨识效率。利用电力科学研究院的综合程序进行了大量的仿真试验验证,结果表明所提方法可以有效解决等值模型多解问题,且辨识精度与辨识效率也显著地得到了提高。 (3)深入研究了基于Volterra级数的建模方法,分析了水轮发电机组这类旋转机械系统的激励力与振动的关系,建立了系统的Volterra模型。深入分析旋转机械系统Volterra时域模型输入量的特点及传统辨识方法的不足,提出了一种改进的Volterra级数模型辨识方法,提高了Volterra时域模型的辨识精度。同时,依据Volterra核函数能够反映系统结构特性的特点,构建了基于神经网络与Volterra时域核函数的故障诊断模型,并通过仿真算例和实验分析对所提方法进行了验证。 (4)针对水轮发电机组转轮状态监测与识别困难的问题,提出运用基于高阶统计量的盲辨识方法对水轮机转轮时域Volterra模型进行参数辨识。同时,探讨了Volterra模型的频域形式,构建了水轮机转轮的广义频率响应模型,通过水轮机转轮广义频率响应分析了其工况的变化。 (5)针对水轮发电机组故障样本不足而限制了其故障诊断方法发展的问题,提出运用有限元仿真方法对水轮发电机组故障振动响应进行建模仿真,并提出了一种基于在线测量信息的水轮发电机组非线性输出响应函数辨识方法。利用非线性输出函数特征量与SVM分类器构建了水轮发电机组故障诊断系统,实现了水轮发电机组振动故障的有效识别。
[Abstract]:With the continuous development of the hydropower industry in China, large hydropower stations have been built and put into operation. The turbine generator sets gradually to large capacity and high water head direction. It is important to ensure the safe and stable operation of the turbine generator set to improve the economic benefits of the power plant and ensure the safety of the power plant and the basin. The operation mechanism of the strong nonlinear and complex giant systems with multi field coupling is not completely clear. The modeling and fault diagnosis of the unit is always the difficult problem in the related research and engineering application. In this context, the system identification theory of the turbine generator set is deeply studied, the modeling of the turbine generator set and the model parameter identification research are carried out. It not only has important theoretical significance and engineering application value for improving the control quality of the turbine generator set, improving the power quality and maintaining the stability of the power system, but also provides theoretical basis and technical support for the fault diagnosis of the turbine generator set. This paper is based on the model parameter identification of the turbine generator set and the fine modeling of the unit. As well as the problems in fault diagnosis, in order to establish a fine turbine generator model and explore advanced fault diagnosis methods, through in-depth analysis of the actual operation characteristics of hydroelectric units, combined with system identification theory, intelligent optimization method and dynamic system identification method, it is suitable for unit control and power system analysis. The model based fault diagnosis method of water turbine generator set is carried out in depth. The modeling theory of Volterra series, generalized frequency response function (GFRF) and nonlinear output frequency response function is introduced, and the model of turbine generator set is established by the method of parameter identification, and the model is used to model the model. The main research contents and innovative achievements of this paper are as follows:1.
(1) the existing hydraulic turbine model is deeply studied, and the applicable scope and advantages and disadvantages of various hydraulic turbine models are summarized and summed up. In view of the problem that the hydraulic loss of the water guide mechanism can not be ignored and the model mechanism is complex, the method of curve fitting is introduced to establish the fine model of the hydraulic turbine which consider the hydraulic loss of the water guide mechanism. It is introduced to the parameter identification of the fine model of the turbine. In order to improve the slow convergence rate and easy to fall into the "local optimum", an improved artificial fish swarm algorithm (IAFSA) is proposed to improve the convergence and accuracy of the algorithm. The proposed method realizes the hydraulic loss curve of the hydraulic turbine refinement model and the water guide mechanism. Step identification.
(2) in order to simplify the network structure of the power system and meet the needs of the analysis of the power system, the equivalent modeling of the small and medium hydropower units is studied. The equivalent model of the five order generator set is set up more suitable for the salient effect of the turbine generator, and a small and medium hydropower cluster based on the study of the data of the multi point PMU measurement data and the contact line data is proposed. The target function is identified by the equivalent model, and the identification strategy is improved to improve the identification efficiency. A large number of simulation tests are carried out by the comprehensive program of the Institute of power science. The results show that the proposed method can effectively solve the problem of multi solution of the equivalent model, and the identification accuracy and identification efficiency are also greatly improved.
(3) the modeling method based on Volterra series is deeply studied. The relationship between the excitation force and the vibration of the rotating machinery system of the turbine generator set is analyzed, and the Volterra model of the system is established. The characteristics of the input quantity of the Volterra time domain model of the rotating machinery system and the shortcomings of the traditional identification method are deeply analyzed, and an improved Volterr is put forward. The identification method of the A-number model improves the identification precision of the Volterra time domain model. At the same time, the fault diagnosis model based on the neural network and the time domain kernel function of the Volterra is constructed according to the characteristics of the Volterra kernel function which can reflect the structure characteristics of the system, and the proposed method is verified by the simulation example and the experimental analysis.
(4) aiming at the problem of the difficulty in monitoring and identifying the state of the turbine runner, a blind identification method based on high order statistics is used to identify the parameters of the time domain Volterra model of the turbine runner. At the same time, the frequency domain form of the Volterra model is discussed, the generalized frequency response model of the turbine wheel is constructed, and the turbine runner is constructed. The generalized frequency response analysis is used to analyze the change of its working condition.
(5) in view of the shortage of fault samples of water turbine generator sets, the problem of the development of the fault diagnosis method is limited. The finite element simulation method is used to model and simulate the fault vibration response of the turbine generator set, and a nonlinear output response function identification method based on the on-line measurement information is proposed. The fault diagnosis system of hydro-generator set is constructed based on linear output function and SVM classifier, and the effective identification of vibration fault of hydro-generator set is realized.
【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TV734.1;TV738

【相似文献】