动态无功优化的混合智能算法

Hybrid Intelligent Method for Dynamic Reactive Power Optimization

针对存在离散控制设备动作次数约束的动态无功优化问题,提出免疫遗传算法和非线性内点法的混合算法.首先忽略控制设备的离散性和动作次数约束,采用非线性内点法求解初始优化解;然后按照控制变量的性质将原问题分解为连续优化与离散优化2个子问题迭代求解.在离散优化问题中,保持连续变量不变,采用免疫遗传算法优化离散变量,通过特别的编码方式使抗体自动满足动作次数约束;在连续优化问题中,保持离散变量不变,采用非线性内点法优化连续变量.混合算法充分结合了免疫遗传算法和非线性内点法的优点,能较快求解动态无功优化的近似最优解.IEEE14节点系统的仿真结果验证了混合算法的有效性.

In dynamic reactive power optimization problem （DRPO）, action number constraints of discrete variables should be considered. By integrating immune genetic algorithm （IGA） and nonlinear interior point method （NIPM）, a hybrid method for DRPO is proposed. First, the original DRPO problem is converted to a continuous optimization problem by relaxing the discrete variables, and the solution is obtained by NIPM. Then, according to the feature of control variables, the original DRPO problem is decomposed into a continuous optimization sub-problem and a discrete optimization sub-problem. The discrete optimization sub-problem is solved by IGA, and the continuous one is solved by NIPM. By solving the two sub-problems alternately, the optimal solution of the DRPO can be obtained. The proposed hybrid method combines advantages of IGA and NIPM, and finds the approximate optimal solution of DRPO. Numerical simulations on the IEEE 14 bus system illustrate that the proposed hybrid method is effective.