基于信赖域序列二次规划算法的含DG配电网优化控制

Optimal Control of Distribution Network Containing Distributed Generations Based on Trust Region Sequential Quadratic Programming Algorithm

针对含分布式电源(distributed generation,DG)配电网运行中的DG优化控制问题,建立了包括最大化DG出力、最小化系统网损和电压偏差3个目标的数学模型,提出一种信赖域序列二次规划(trust region sequential quadratic programming,TRSQP)算法来求解DG的最优化有功和无功出力。该算法在信赖域的指引下依照一定的规则,通过形成一系列信赖域内二次规划子问题(QP)来逐步求解原问题。通过灵敏度分析方法减小了二次规划子问题的规模并采用有效集法求解。以修改过的IEEE 33节点为例,对最优化DG出力问题进行了仿真,结果表明提出的最优化DG出力模型和求解算法能够保持网络运行在安全经济的状态;以33、69、292、588、1180节点配电系统为例,将文中算法与逐次线性规划法(sequential linear programming method,SLPM)和原对偶内点法(primal-dual interior-point method,PDIPM)进行了比较,结果表明所提算法具有较快的计算速度和良好的鲁棒性,适合在线应用。

A mathematical model containing three objectives of maximized output of distributed generations （DG）, minimized network loss and minimized voltage deviation is established for optimal control of DG during the operation of distribution network containing DG, and a trust region sequential quadratic programming （TRSQP） algorithm is proposed to solve the optimization of DG and reactive power output. Under the guidance of trust region and according to certain rules, by means of forming a series of sub-problems of quadratic programming within the trust region the primal problem is solved step-by-step. Using sensitivity analysis the scale of the sub-problem of quadratic programming is decreased and solved by active set method. Taking the modified IEEE 33-bus system as example, the output of optimized DG is simulated, and simulation results show that the proposed optimized DG output model and the solving algorithm can make the distribution network kept secure and economic operation condition; taking 33-bus, 69-bus, 292-bus, 588-bus and 1180-bus distribution networks respectively for examples, the proposed algorithm is compared with sequential linear programming method （SLPM） and primal-dual interior-point method （PDIPM） respectively, and comparison results show that the proposed algorithm possesses good robustness and its computing speed is faster, so it is suitable for on-line application.