基于可观性分析和多步预报CKF的双馈风机动态状态估计

Dynamic State Estimation of Doubly-fed Wind Turbine Based on Observability Analysis and Multi-step Prediction Cubature Kalman Filter

准确的动态状态估计结果对实时监测双馈风机运行状态具有重要意义。适当的量测集选取是实现准确状态估计的基础。为此,基于Lie导数推导了系统状态可观性矩阵;然后,基于自动微分技术实现了可观性矩阵的快速计算,并利用可观性矩阵的最小奇异值来定量评价不同量测集的可观性。双馈风机受到扰动后状态变化迅速,考虑到在卡尔曼滤波的预报步中不使用量测值,提出了基于多步预报的容积卡尔曼滤波(CKF)方法。利用改进IEEE 39节点系统对双馈风机进行了可观性分析,选出了使双馈风机状态具有较高可观性的量测集,并在此基础上验证了所提多步预报CKF方法的有效性。

Accurate dynamic state estimation results are of great significance for monitoring the operating states of doubly-fed wind turbines in real time. Appropriate measurement set selection is the foundation for the realization of accurate state estimation.Therefore, the observability matrix of nonlinear system states is derived based on the Lie derivative, and the fast calculation of the observability matrix is implemented based on the automatic differential technique. The smallest singular value of the observability matrix is used to quantitatively evaluate the observability of different measurement sets. Considering that the prediction step of the Kalman filter does not use the measurement value, a cubature Kalman filter(CKF) method based on adaptive multi-step prediction is proposed. The observability analysis of the doubly-fed wind turbines is carried out by using the modified IEEE 39-bus system,and the measurement set is selected to make the states of doubly-fed wind turbines have high observability. On this basis, the effectiveness of the proposed cubature Kalman filter method based on multi-step prediction is verified.