【作者】 刘兰；

【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】 随着我国电力系统的迅速发展,电力系统的结构和运行方式日趋复杂,电力系统调度中心的自动化水平也正在由低级向高级发展。在现代的调度系统中,计算机已成为重要的一环,其高级自动化功能主要体现在它所具备的程序功能,因此,状态估计算法的可靠性是我们必需关注的问题。 目前的状态估计算法大多是沿用经典的加权最小二乘估计法,它的前提是建立在随机噪声为正态分布的基础上,当出现粗差时,加权最小二乘法必须结合其它不良数据检测和辨识方法才能得到较好的估计结果,由于量测噪声不可能严格服从正态分布,结合其它检测辨识方法剔除不良数据增加了程序的复杂性,延长了算法的计算时间。针对这些问题,本论文提出将抗差思想应用于电力系统状态估计中,应用抗差最小二乘法对电力系统进行状态估计,并将估计结果与同等环境下的加权最小二乘法的估计结果进行比较,证明抗差最小二乘法的抗差效果良好,效率比加权最小二乘估计法高。 选取不同的等价权函数,其相应的抗差能力也不一样。用已有的huber法和IGG法的等价权函数进行验算时,相同情况下,huber法比IGG法的抗差效果要好。从这一情况出发,本文提出了几种新的方案,在等价权函数的选取原则下构造了三种新的等价权函数,并将这三种新方案在具体算例中进行状态估计验算,结果表明:不同的等价权函数都具有一定的抗差能力,但它们的抗差效果不同,方案一、三的效果较huber法好,方案二的效果较huber法差。由此可见,选取好的等价权函数也是提高状态估计算法效率的重要途径之一。 估计的抗差性是通过在估计计算过程中变权实现的,迭代修正方程的系数矩阵随着量测噪声的情况而变化,有可能导致数值解的不稳定,因此,本文将正交变换法用于抗差最小二乘法中,并在实例中进行验算,结果表明:算法的状态估计数值解的稳定性良好。更多还原

【Abstract】 Along with the development of power systems in our country, the structure and run-mode of which are more and more complex, and the automation level in the control center is also developing from base to advance. To the computer, which plays an important part in modern control center, the advanced automation control depend on the functions of the program installed, so we should pay sufficient attention to the reliability of state estimation algorithm.At present the most of state estimation algorithms in use follow the classic weighted least squares (WLS), whose precondition is that random noises submit normal distribution. When bad data exists, WLS can not get better estimation results until it combined with other bad data detection and identification methods. Due to that measure noise can’ t strict submit normal distribution, integrated with other detection and identification methods makes the program more complex, and prolongs the computing time of the algorithm. To this problem, the idea applying robust least square estimation to state estimation of power system was proposed. Compared with WLS, it indicates that robust least square is effective and of better efficiency.Different equivalent weight function has different abilities to resist gross. Considering that in the same condition, huber transform works better than IGG transform, some new ways was put forward. Three new equivalent weight functions was built based on the rule of choosing equivalent weight functions. After that, the new schemes were put in use in the experiments of state estimation. The results show that different equivalent weight functions are all able to resist gross, but their ability of resisting gross are different. Scheme. 1 and Scheme. 3 are better than huber transform, on the contrary Scheme. 2 is worse. So we can safely draw conclusion that choosing a good equivalent weight function is an important way to improve the efficiency of state estimation arithmetic too.The implementations of gross resistance capability of estimation rely on the changes of weight during computing. The coefficient matrix of iterative amending equation changes with measurement noise, and it may cause the instability of estimation results. So the proposal of using orthogonal transformation to robust least square was detailed. Experiment results show that numerical solutions of state estimation are of better stability.state estimation;robust least square;equivalent weight function;orthogonal transformation更多还原