基于GMRES的改进连续潮流算法研究

Research on improved continuation power flow with GMRES

阐述了一种基于GMRES的改进连续潮流算法,主要针对大型电力系统,在预测环节中采用拉格朗日非线性预测,减少了计算时间;在校正环节中将GMRES方法与牛顿法相结合构成内外双层迭代,并应用ILU分解对系数矩阵进行预处理,从而避免了对修正方程进行直接求解,提高了连续潮流法的计算速度。该算法根据梯度参数变化对步长进行了有效的控制,可以大大提高PV曲线的追踪效率。采用该算法对IEEE300节点测试系统进行仿真计算,取得了良好的计算结果,从而验证了该方法的有效性和快速性。

An improved continuation power flow is presented based on the generalized minimal residual （G MRES） method.The new method is mainly used in large power systems.Nonlinear predictor with Lagrange quadric interpolation is used in the predictor which is more efficient.In the corrector,double-layer iterative method is composed by Newton method and GMRES method and ILU decomposition is used in the preconditioner.The new method avoids the direct solving of the correction equation and improves the computation speed of continuation power flow.The new method controls step-length according to the change of gradient parameter and can improve the tracking efficiency of PV curves greatly.The new method is used on the IEEE 300-bus test system and the calculation result is good,which demonstrates the validity and rapidity of the proposed continuation power flow method.