【作者】 贺觅知；

【导师】 黄彦全；

【作者基本信息】 西南交通大学 ， 电力系统及其自动化， 2006， 硕士

【摘要】 电力系统状态估计是电力系统调度、控制、安全评估等方面的基础，也是能量管理系统(EMS)的核心组成部分。动态状态估计通过状态预测能对电力系统进行安全评估，实现经济调度、预防控制等在线功能，重要性不言而喻，因此本文对基于扩展卡尔曼滤波的电力系统动态状态估计进行了分析研究。 本文首先通过对卡尔曼滤波算法的过程及计算公式的推导，阐述了电力系统模型下扩展卡尔曼滤波动态状态估计器的工作原理和实际应用中所面临的问题，并列举分析针对这些问题的几种改进算法。 在扩展卡尔曼滤波算法的基础上，本文提出一种自适应卡尔曼滤波原理的动态状态估计新算法，该算法通过采取增广最小二乘法来修正状态模型参数和指数加权法来估计时变噪声。这样，在利用量测数据进行状态估计的同时，来不断的对系统模型参数和噪声统计特性进行估计和修正以提高模型精确度，降低状态估计的误差。此外，本文算法还采取三次样条插值的数值分析方法，利用状态估计的历史数据来拟合状态函数，估计出状态转移矩阵的初值，以缩短状态估计周期和提高算法精确度；同时采取稀疏矩阵技术以减少算法的计算量。 仿真试验结果表明：本文算法不仅在电力系统正常状况下具有较好的预测、估计结果，而且在电力系统发生负荷突变和存在量测不良数据的异常情况下，也体现出良好的适应性。更多还原

【Abstract】 Power system state estimation is not only the foundation of power system dispatch, control and security assessment, but also the core of Energy Management System. Through state prediction, dynamic state estimation is able to finish the security assessment in order to achieve economic dispatch, prevention control and other online functions. Therefore researches on power system dynamic state estimation based on extended Kalman filter were carried outAlong with the derivation of Kalman filter algorithm, the operating principle of extended Kalman filtering dynamic estimator in power system and the problems facing practical application were detailed. Then several improved algorithms were proposed.On the basis of extended Kalman filter algorithm, a new dynamic state estimation algorithm based on adaptive Kalman filter was brought up. During the process of state estimation, recursive extended least square was introduced to update the parameters of state model, at the same time exponential weighted algorithm was put in use to estimate the statistical characteristic of time-varying noise. Thus the errors of state estimation could be reduced. What’s more, cubic spline interpolation was proposed to fit the state function using historical state data. After that, the initial value of state transfer matrix could be estimated. All of that improvements together with sparse matrix technique contribute to the cut of calculation period and the increase of accuracy.The simulation results show that with proposed algorithm, better prediction and estimate results could be got under normal condition in power system, further more when sudden changes or bad data were involved, this algorithm could still work with nice adaptability.更多还原