基于BP神经网络的城网供电可靠性预测方法

A BP Neural Network Based Method to Predict Power Supply Reliability of Urban Power Network

传统的供电可靠性评估方法是以准确的配电网结构和多年的元件可靠性历史数据为基础的,难以实现城市复杂配电网远期供电可靠性指标的预测。为此文章提出一种基于BP神经网络的城市电网供电可靠性预测方法,首先找出影响供电可靠性指标的几个主要特征量,包括最大负荷、架空线平均长度、线上平均分段开关台数、线上平均联络开关台数、线路平均配变台数和线路平均配变容量,将这些特征量的历史数据作为输入样本对人工神经网络进行训练,利用训练好的网络就可以预测规划目标年的城市电网供电可靠性指标。对某城市电网的应用结果表明该方法是有效的,所采用的BP神经网络具有较好的收敛性。通过对影响供电可靠性的相关因素进行灵敏度分析还可以获得对供电可靠性指标较敏感的相关特征量,供电企业可以据此制定提高可靠性的相关措施。

Due to the fact that traditional power supply reliability evaluation is on the basis of true structure of distribution network and multi-year historical data of element reliability, so it is hard to predict long-term power supply reliability of complex urban distribution network. For this reason, the authors propose a BP neural network based method to predict power supply reliability of urban power network. Firstly, several principal characteristic quantities impacting power supply reliability indices, such as peak load, average length of overhead lines, average quantity of sectional switch of the lines, average quantity of tie switch of the lines, average quantity of distribution transformers of the lines and average capacity of distribution transformers of the line, are found out; then taking historical data of these principal quantities as input samples, the artificial neural network （ANN） is trained; finally, using the trained ANN, the power supply reliability indices of urban power network in target year can be predicted. The results of applying the proposed method to a certain urban power network show that the proposed method is effective and the adopted BP neural network possesses good convergence. By means of sensitivity analysis on relevant factors impacting power supply reliability, relevant characteristic quantities sensitive to power supply reliability indices can be obtained, on this basis power supply companies can draw up related measures to improve power supply reliability.