基于改进聚类和RBF神经网络的台区电网线损计算研究

电网线损是反映低压台区配电网输配电性能好坏的重要技术指标，针对低压台区线损计算不精确和效率不高的问题提出了一种有效的计算方法，即基于改进K-Means聚类和正交最小二乘法（OLS）优化的径向基（RBF）神经网络计算模型。首先通过层次分析法（AHP）对线损的电气指标进行提取，根据得到的线损指标用改进的K-Means聚类算法进行分类处理，然后用OLS改进的RBF网络和标准的RBF网络分别对得到的分类样本进行训练，再用训练好的模型计算台区电网线损。最后利用某地区低压台区的68组样本验证了所提方法的有效性。

Power Line Loss Calculation in Low Voltage Region Based on Improved Clustering Algorithm and RBF Neural Network

Power line loss is an important index reflecting the performance of transmission and distribution in low voltage region. This paper proposes an effective calculation method to solve the problem of the inaccurate and inefficient calculation of network loss in the low voltage region， which is based on improved K-means clustering and orthogonal least squares （OLS） optimized radial basis function （RBF） neural network calculation model. Firstly， analytic hierarchy process（AHP）is used to extract the electric indices for the line loss， and the indices are classified by the modified K-means clustering algorithm. Then the OLS-optimized RBF network and standard RBF network is used respectively to train the classification samples，and the trained model is used to calculate the line loss in the low voltage region. Finally， a numerical simulation is carried out with 68 groups of samples from a certain low voltage region to validate the effectiveness of method.