摘要
为了准确计算低压台区配电网的线损,提出遗传算法优化BP神经网络的配电网理论线损计算。通过电网线路获取线路特征参数,对特征参数建立BP神经网络的预测模型;根据BP神经网络的拓扑结构中的权值和阈值来确定染色体长度;通过遗传算法来确定BP神经网络的参数,拟合影响线损的特征参数和线损的复杂关系。该算法减少了线损计算误差,计算时间少,稳定性增强。
In order to accurately calculate the line loss of distribution network in low-voltage stations, we proposed the theoretical line loss calculation of distribution network of optimized BP neural network based on genetic algorithm. Line characteristic parameters were obtained through network lines, and BP neural network prediction model for characteristic parameters was established. We determined the chromosome length according to the weights and thresholds in the topological structure of BP neural network. The parameters of BP neural network were determined by genetic algorithm, and the complex relationship between the characteristic parameters and the line loss was fitted. The algorithm reduces the calculation error of line loss, reduces the calculation time and enhances the stability.
引文
[1] 薛禹群,朱学愚.地下水动力学[M].北京:地质出版社,1979.
[2] 杨秀台.电力网线损的理论计算和分析[M].北京:水利电力出版社,1985
[3] 李桂鑫. 基于遗传算法的RBF神经网络用于配电网线损计算[D]. 天津:天津大学, 2007.
[4] 李勇平. 基于改进粒子群神经网络的电信业务预测模型研究[D]. 广州:华南理工大学, 2009.
[5] 吴官安,倪保珊.电力系统线损[M].北京:中国电力出版社,1996.
[6] 王小川,史峰,郁磊,等.MATLAB神经网络43个案例分析[M].北京:北京航空航天大学出版社,2013.
[7] 文福栓,韩祯祥.基于分群算法和人工神经网络的配电网线损计算[J].中国电机工程学报,1993,13(3):41-51.
[8] 文福拴, 韩祯祥. 联合采用Kohonen模型和BP模型的配电网线损计算[J]. 电工技术学报, 1993(3):16-20.
[9] 张晓缋, 戴冠中, 徐乃平. 一种新的优化搜索算法——遗传算法[J]. 控制理论与应用, 1995(3):265-273.
[10] 恽为民,席裕庚.简单遗传算法的运行机理研究[C]//中国自动化学会第四届全国学术年会,1993:29-36.
[11] Yen G G, Lu H. Hierarchical genetic algorithm based neural network design[C]//2000 IEEE Symposium on Combinations of Evolutionary Computation and Neural Networks.IEEE, 2000.