电网无功优化算法的研究与实现

发布时间：2018-11-05 11:58

【摘要】：随着电力网结构日趋稠密复杂以及电网智能化程度的日益加深,电网调度系统要求更高效且更易于操作的电网无功优化计算方法。因此,如何进一步增强电网的调控能力,提高无功优化的效率,成为电力系统研究者关注的问题。电压无功优化控制作为电力网调度的基本任务,是通过调度系统自动化地采集场站节点信息来进行实时数据地综合分析与控制,最终达到使电压的合格指数更高,电容器投切更加科学合理以及电网网损更低的目标。现有的无功优化模型有:单一目标模型、多重目标动态优化模型等;优化算法包括线性规划法、简化梯度法等传统算法以及遗传算法、模拟退火算法等智能算法。在实际应用中,规划类算法虽然有严格的理论支持,但较难处理具有大量离散变量和约束条件的问题,并且复杂无功优化模型的实时控制达不到要求。而对于单一智能算法在优化计算时来说,算法易早熟,易陷入整个解空间内局部最优的陷阱,并且存在计算时间长,后期寻优速度慢等缺陷。因此,针对无功优化在实际应用中遇到的难点,本文在综合分析了属于无功优化范畴内的科研状况之后,研究设计出变惯性权值和加速因子的粒子群算法。在改进算法的更新公式中惯性权重w和加速因子c2会根据粒子与最优粒子之间的距离来更改w和c2的值,比如当粒子靠近群体最优粒子时,增大惯性权重w,减小加速因子c2。同时,将算法与遗传算法、模拟退火算法和蚁群算法进行比较,试验对比发现算法在降低电网损耗方面效果较为明显,且算法优化时间较短,在全局寻优的把握较大。最后将改进后的算法应用到基于struts2框架和oracle数据库开发出来的四川自贡地区电网无功优化可视化软件系统中。
[Abstract]:With the increasing density and complexity of power grid structure and the deepening of power network intelligence, power dispatching system requires more efficient and easy to operate reactive power optimization calculation method. Therefore, how to further enhance the power grid control ability and improve the efficiency of reactive power optimization, become the focus of power system researchers. As the basic task of power grid dispatching, the voltage and reactive power optimal control is to collect the node information of the station automatically to analyze and control the real-time data, and finally to make the voltage quality index higher. The target of capacitor switching is more scientific and reasonable, and the power loss is lower. The existing reactive power optimization models include: single objective model, multi-objective dynamic optimization model, optimization algorithm including linear programming, simplified gradient method and other traditional algorithms, genetic algorithm, simulated annealing algorithm and other intelligent algorithms. In practical application, although the programming algorithm has strict theoretical support, it is difficult to deal with the problems with a large number of discrete variables and constraints, and the real-time control of the complex reactive power optimization model can not meet the requirements. For the single intelligent algorithm, it is easy to precocious and fall into the trap of local optimum in the whole solution space, and it has some defects such as long calculation time and slow searching speed. Therefore, aiming at the difficulties encountered in the practical application of reactive power optimization, this paper studies and designs a particle swarm optimization algorithm with variable inertia weight and acceleration factor after synthetically analyzing the scientific research situation in the category of reactive power optimization. In the updating formula of the improved algorithm, the inertial weight w and the acceleration factor c2 will be changed according to the distance between the particle and the optimal particle. For example, when the particle is close to the swarm optimal particle, the inertia weight w will be increased and the acceleration factor c2 will be reduced. At the same time, comparing the algorithm with genetic algorithm, simulated annealing algorithm and ant colony algorithm, it is found that the algorithm is more effective in reducing power loss, and the optimization time is shorter and the global optimization is better. Finally, the improved algorithm is applied to the visual software system of reactive power optimization in Zigong area based on struts2 framework and oracle database.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM714.3

【参考文献】