改进粒子群-禁忌搜索算法在多目标无功优化中的应用

Application of improved particle swarm-tabu search algorithm in multi-objective reactive power optimization

针对有功网损、电压偏差和静态电压稳定裕度的多目标无功优化问题,提出一种基于改进粒子群-禁忌搜索算法的多目标电力系统无功优化方法。以最小特征值模为电压稳定裕度指标建立了3个目标函数的单一妥协模型。应用Kent映射产生的混沌序列作为初始种群,保证初始种群的多样性和均匀性。粒子群优化(PSO)算法进行前期计算时,采用凸函数递减惯性权重和自适应学习因子提高算法的收敛速度和精度;针对PSO算法搜索精度不高和陷入局部最优的问题,在PSO算法后期收敛后引入禁忌搜索算法全局寻优。基于群体适应度方差,引入模糊截集理论将模糊集合转化为经典集合,定义了经典集合下的收敛指标,当其值为0时进入禁忌搜索计算阶段,解决2种算法的切换问题。将所提方法应用于IEEE14、IEEE30和IEEE118节点系统中,验证了其有效性和可行性。

A method based on the improved particle swarm-tabu search algorithm is proposed for the reactive power optimization of power system with three objectives：active power loss ,voltage deviation and static voltage stability margin. With the minimum eigenvalue modulus as the voltage stability margin index,a single-compromise model including three objective functions is built. The chaotic sequence produced by Kent mapping is taken as the initial population to insure its diversity and uniformity. The convex decreasing inertia weight and the adaptive learning factor are adopted in the preliminary calculation of PSO（Particle Swarm Optimization） to improve its convergence speed and accuracy,while the tabu search is applied in the post-convergence calculation of PSO to avoid the low search accuracy and local optimization. According to the fuzzy cut-set theory and based on the variance of population fitness, the fuzzy set is converted into the classic set,for which a convergence indicator is defined. The tabu search starts only when the indicator value is 0. The proposed method is applied to IEEE 14-bus, IEEE 30-bus and IEEE 118-bus systems for verifying its effectiveness and feasibility.