论文部分内容阅读
以内点法求解最优潮流(optimal power flow，OPF)的经典非线性规划模型已得到广泛应用，但无法保证解的全局最优性。而求解 OPF 的半正定规划模型，在一定条件下能获得全局最优解，但存在计算时间长和可能无法获得可行解的缺点。因此，文中提出一种结合非线性规划和半正定规划模型两者优势求解 OPF 问题的混合优化方法，以实现在更短的时间内获得全局最优解。首先，提出验证由内点法求解 OPF非线性规划模型(nonlinear programming，NLP)所得解是否为全局最优的充分条件。若非全局最优，则基于 OPF的半正定规划模型给出由该局部最优解出发的下降方向，并通过步长控制得到新的初值，交由内点法重新求解 OPF 的非线性规划模型。算例测试结果表明，该算法在避免求解完整半正定模型需耗费大量时间的同时，能够有效跳出非线性规划模型的局部最优解，收敛到全局最优解或更优的解。“,”The interior point method for OPF in a nonlinear programming (NLP), which has been widely applied, can’t guarantee a global optimum due to the non-convexity of OPF. Solving OPF in semidefinite programming (SDP), which is convex, a global optimum solution can be found when the relaxation condition is satisfied. However, the computational time will increase rapidly when SDP applied to large-scale system andthe relaxation condition is not always satisfied, which means an infeasible solution of the original OPF problem may be obtained. Therefore, this paper presented a new hybrid iterative algorithm by combiningthe semidefinite programming and nonlinear programming, which could find the global optimum solution during a shorter time than SDP. At first, we developed a SDP-inspired sufficient condition for global optimality of a candidate solution of NLP model. When candidate solution was not global optimal, we calculated another initial point for NLP model by using a descend direction based on SDP model, aiming to escape from the neighborhood of local minima and converge to a better or global solution. The numerical results show that the proposed algorithm can escape from local minima and converge to a better or global solution.