多随机激励下风电机组在线辨识建模研究

项目名称： 多随机激励下风电机组在线辨识建模研究

项目编号： No.51207045

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 电气科学与工程学科

项目作者： 潘学萍

作者单位： 河海大学

项目金额： 25万元

中文摘要： 本项目以在线辨识风电机组各组成模块的模型参数为目标，以构建有效的参数辨识算法为基础，研究多随机激励下，风电机组的驱动系统、风力发电机、风电控制系统的参数是否可以辨识、如何辨识的机理。根据激励信号的时间尺度及风电机组快变/慢变动态特性的对应关系，将受扰轨线进行多尺度分解，解耦辨识与快变/慢变动态对应的模型参数；基于此，分析多随机激励下风电机组参数的可辨识性；提出综合考虑频域灵敏度与激励信号强度的功率谱灵敏度指标，分析风电机组参数辨识的难易度；采用2步辨识的思路，先将风电机组模型参数在时间尺度上解耦辨识，再协调优化获得风电机组的所有参数；以大扰动激励下的参数辨识结果为基准，校核并修正小扰动激励下的辨识值。本研究将确立符合实际运行工况的风电机组模型参数，为大规模风电场并网研究提供有效的参数。

中文关键词： 风电机组；建模；参数辨识；模型验证；模型误差

英文摘要： This proposal aims to identify model parameters for wind turbine generators(WTGs) by online identification algorithms. It studies the model parameter identifiability and how to identify them for each component (wind turbine, generator/converter,converter control) of WTGs following multiple stochastic excitations. Firstly, the measured ambient data excited by wind turbulence and load random variations are decomposed into different time-scale components, with the purpose of making the fast and slow model parameters to be identified separately. Secondly, the parameter identifiability of WTGs is studied. Thirdly, the sensitivity index, which takes into account both frequency-domain sensitivity and power spectral density of input signals is proposed to measure the difficulty of identification, which means the possibility of obtaining the accurate values for the parameters. Fourthly, the two step identification process is presented: (1) use the stochastic subspace identification (SSI) to estimate the parameters corresponding to fast/slow dynamic performance; (2)apply the coordinate optimization technique to get all parameters. Lastly, the identified results following severe disturbances are used as references to validate the parameters identified from ambient data. This project attempts to attain accurate representat

英文关键词： wind turbine generator；modelling；parameter estimation；model validation；model error