基于专家样本库和最小二乘支持向量机的配电网线损率预测模型

配电网线损率是评价电网企业运营的一项重要经济指标,利用电力运行数据建立线损率模型时建模样本对模型的预测精度具有重要影响。对此,提出了一种基于专家样本库和最小二乘支持向量机(LS-SVM)的线损率计算模型,利用离散粒子群优化算法从台区配电网电力运行数据中寻找包含较大运行状态信息量的样本记录,构建专家样本库,考虑台区配电网的有功供电量、无功供电量、端口电流、居民容量占比、表计数目及气象温度和气象湿度,基于专家样本库采用LS-SVM算法建立配电网线损率预测模型,并利用该模型对0.4kV配电网线损率进行计算验证。结果表明,基于专家样本库建立的线损率模型与普通样本建立的模型相比,具有更高的预测精度。

Prediction Model of Line Loss Rate in Distribution Systems Based on Expert Sample Base and LS-SVM

The line loss rate in distribution systems is an important economic indicator for evaluating grid enterprise operations. When using the power operation data to establish the line loss rate model, the selected samples have an important impact on the prediction accuracy of the model. Aiming at this problem, a line loss rate prediction model was proposed based on expert sample base and least squares support vector machine (LS-SVM). The discrete particle swarm optimization algorithm was applied to find the data that contain maximum operating condition information from the operating data of multiple area of the grids. The expert sample base was constructed based on the selected data. Considering the active power generation, the reactive power generation, port current, residential capacity ratio, number of electricity meters, meteorological temperature and meteorological humidity, the line loss rate prediction model was established using the LS-SVM method. Based on the model, the line loss rate of 0.4 kV power distribution network was calculated and verified. Compared with the model developed with common samples, the experimental results show that the line loss rate calculation model developed with the expert sample base has better prediction accuracy.