基于改进弹性网络回归的电容式电压互感器误差状态预测

Error state Prediction of Capacitive Voltage Transformer Based on Improved Elastic Network Regression

为了探索各因素对电容式电压互感器(Capacitance Type Voltage Transformer,CVT)误差状态的影响程度,提高CVT误差状态预测的精度,提出一种基于改进弹性网络回归的CVT误差状态预测方法。首先,通过拉依达准则和K-最近邻填充方法对原始数据进行预处理;然后,构建改进的弹性网络回归模型,根据每个影响因素的皮尔逊相关系数在弹性网络回归模型中给与不同的惩罚权重,计算出该模型的最优系数;最后,利用具有最优系数的回归模型对CVT误差状态进行预测。结果表明所提方法具有更高的预测精度。

The error state of capacitive voltage transformer(CVT)is coupled by temperature and humidity,electric field,secondary load and other factors,which leads to poor stability and reliability,and then affects the fairness of power trade settlement and the safe operation level of power grid.In order to explore the influence of various factors on CVT error state and improve the accuracy of CVT error state prediction,a method of CVT error state prediction based on improved elastic network regression was proposed.Firstly,the raw data are preprocessed by the Laida criterion and K-nearest neighbor filling method.Then,an improved elastic network regression model is constructed,and different penalty weights are given to the elastic network regression model according to the Pearson correlation coefficient of each influencing factor,and the optimal coefficient of the model is calculated.Finally,the regression model with optimal coefficient is used to predict the error state of CVT.The ablation experiment and comparison experiment with the data collected from Xiashi Substation in Taizhou show that the proposed method has higher prediction accuracy.