摘要
对地区配电网空间负荷的预测,能够高效规划配电网资源。对配电网空间负荷的优化预测,需要采用决策偏好概率函数决定预测的全局最优解,完成空间负荷预测。传统方法结合模糊决策的方案,利用隶属度函数来定义空间负荷的地域特点,但忽略了求取预测的全局最优解,导致预测精度偏低。为此,提出基于改进粒子群算法,针对负荷密度分布、用地类型动态特征以及空间小区动态划分等多个目标决策因素对空间负荷进行预测,通过建立的分类模型,描述预定的数据类集或者概念集,采用分类规则集的形式来表述,使用分类规则对新的数据集进行划分。将负荷进行分类,再将规划区域分成若干小区后以划分的小区为对象,预测小区的负荷值,采用决策偏好概率函数决定预测的全局最优解。仿真结果表明,与传统的空间负荷预测方法相比,所提方法能够有效提高预测的准确度。
This paper proposes a spatial load prediction algorithm based on improved particle swarm. For decision factors of multiple objectives such as distribution of load density, dynamic characteristics of land and dynamic division of spatial small area, the spatial load is predicted, through classification models, predetermined data set or concept set is described, then classification rule set is used to express and to divide new data sets. The loads are classified, then planning area is divided into some small areas, and divided area is taken as object, and load value of small area is predicted. Finally, probability function of decision preference is used to determine global optimal solution of prediction. Simulation results show that the proposed method can effectively improve the prediction accuracy compared with traditional spatial load prediction methods.
出处
《计算机仿真》
北大核心
2017年第12期125-128,共4页
Computer Simulation
基金
国家自然科学基金(61501185)
北京市自然科学基金(4164101)
河北省自然科学基金(F2016502062)
关键词
地区配电网
空间负荷
优化预测
Regional distribution network
Space load
Optimal prediction
