用遗传算法辅助设计的人工神经网络计算配电网线损

CALCULATION OF LINE LOSSES IN DISTRIBUTION SYSTEMS USING ARTIFICIAL NEURAL NETWORK AIDED BY GENETIC ALGORITHM

利用遗传算法（ＧＡ）辅助设计的人工神经网络（ＡＮＮ）进行配电网线损计算。首先通过ＧＡ来确定ＡＮＮ的三个参数（隐含层中神经元个数、学习速率和动量因子），然后应用ＢＰ型神经网络来拟合影响线损的特征参数与线损之间的复杂关系，在此基础上提出了计算配网线损的简单、准确而实用的方法。实例计算结果表明，该算法比现有其它计算配网线损的方法更为准确。

A method for calculating the line losses in distribution system is presented using Artificial Neural Network(ANN) aided by Genetic Algorithm(GA). The first step is to determine three parameters of ANN, i.e., numbers of neurons in the hidden layer, learning rate and momentum constant. Then BP models are applied to map the complex relationship between line losses and the related parameters. Based on these two steps, a simple, practical and accurate method is put forward. Simulation results for a practical example show that the method is more accurate than other ones.