电力系统状态空间方程的优化降阶研究

发布时间：2018-11-20 05:47

【摘要】：大区域电力系统的互联和系统控制器的多样化、复杂化,导致系统动态方程数急剧增加。“三华”电网中,小干扰稳定分析的状态矩阵的维数已达两万多。求解全维特征值的QR算法具有数值稳定性强的明显优势,但在用于大规模电力系统稳定分析时,舍入误差的影响可能会导致求解失败。本文分析了特定模式降阶法的算法结构和计算过程,特定模式降阶法将状态变量按照特征值灵敏度排序并降阶,采用改进Raleigh商逆迭代计算机电模式的特征值。由于一个降阶模型只能准确计算出一个机电模式特征值,计算效率较低。实际系统中,部分机电模式特性相似或相近,在误差允许范围内可以共用一个降阶模型。本文的主要工作有两个:一是将机电模式分组;二是确定分组内发电机的最优台数。特定模式降阶法的改进。选择发电机组对功角变量的相关因子作为依据,将机电模式分组,每一组共用一个降阶模型,从而大幅减少计算量。降阶模型优化分析。参照一般控制系统的模型降阶方法,确定本文的模型降阶优化思路。通过比较模型中保留的发电机台数不同时,特征值的计算精度,来确定在保证计算精度的前提下,模型中发电机的最优台数。在23机系统和70机系统中进行计算验证。利用相关因子将机电模式分组,参照模型降阶方法的思想优化降阶模型,来计算大规模电力系统机电模式特征值,避免了QR全维特征值法在状态矩阵维数巨大时的求解失败,一个降阶模型可以算出多个机电模式特征值,进一步提高计算效率。
[Abstract]:The interconnection of large area power systems and the diversification and complexity of system controllers lead to a sharp increase in the number of dynamic equations of the system. In Sanhua power network, the dimension of state matrix of small disturbance stability analysis has reached more than 20,000. The QR algorithm for solving full-dimension eigenvalues has obvious advantages of strong numerical stability, but the rounding error may lead to failure in large-scale power system stability analysis. In this paper, the algorithm structure and calculation process of the special mode reduction method are analyzed. The state variables are sorted and reduced according to the sensitivity of the eigenvalue by the special pattern reduction method, and the eigenvalues of the computer electric mode are improved by the inverse iteration of the Raleigh quotient. Because a reduced order model can only accurately calculate the eigenvalue of an electromechanical mode, the computational efficiency is low. In practical systems, some electromechanical modes have similar or similar characteristics, and a reduced order model can be shared within the range of errors. There are two main works in this paper: one is to group the electromechanical mode and the other is to determine the optimal number of generators in the group. Improvement of order reduction method for specific modes. Based on the relative factors of the diagonal variables of generator sets, the electromechanical modes are grouped into groups, each group sharing a reduced order model, thus greatly reducing the amount of calculation. Optimization analysis of reduced order model. According to the model reduction method of the general control system, the optimization idea of model reduction in this paper is determined. The optimal number of generators in the model is determined by comparing the accuracy of the eigenvalues and the number of generators reserved in the model. The calculation verification is carried out in 23 computer system and 70 computer system. By grouping the electromechanical mode with correlation factors and optimizing the reduced-order model according to the idea of the model reduction method, the eigenvalue of the electromechanical mode of large-scale power system is calculated, which avoids the failure of the QR full-dimension eigenvalue method when the dimension of the state matrix is large. A reduced order model can calculate the eigenvalues of multiple electromechanical modes and further improve the computational efficiency.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM712

【相似文献】