集散遗传算法在厂级AGC负荷分配中的应用

Application of Gathering-Scattering Genetic Algorithm to Load Distribution of AGC in Plant Level

针对目前电网采用的自动发电控制(automatic generation control,AGC)方式的缺点,提出了采用发电厂级AGC方式来优化分配厂内各机组负荷,以实现发电成本最小。建立了以各台机组负荷为自变量、发电成本为因变量的多约束条件的厂级AGC负荷分配模型。应用提出的基于种群差异的集散遗传算法进行优化,速度快且精度高。它根据个体间的欧氏距离反映的差异度,通过离散操作提高种群的多样性,通过集合操作提高种群的收敛速度,避免了普通遗传算法早熟和收敛速度慢的缺点。仿真结果表明:基于集散遗传算法的厂级AGC节能效果好,运算时间短且能准确稳定地收敛到最优解。

In view of the defects of the AGC presently applied in power grids, it is proposed to adopt AGC in plant level to implement optimal load distribution among units in the very plant to realize minimum generation cost. A model of load distribution by AGC in plant level is built in which the load of each unit is regarded as independent variable and generation cost as dependent variable, thus this model possesses multi constraint conditions. The load distribution is optimized by population difference based gathering-scattering genetic algorithm （GSGA）, so the optimization process is rapid and accurate. According to the difference degree reflected by Euclidean distances among all individuals, the population diversity is improved by scattering operation and the convergence rate of populations is speeded up by gathering operation to avoid the drawbacks of precocity and low convergence rate of traditional GA. Simulation results show that the energy-saving effect of GSGA based AGC in plant level is satisfied; its operation time is short and GSGA can converge to optimal solution accurately and steadily.