基于径向基神经网络模型的电网地磁感应电流预测

Forecasting of Power Grid GIC Based on RBF Neural Network

为防范地磁感应电流(geomagnetically induced currents,简称GIC)引起的电网无功波动和变压器直流偏磁等影响,采用径向基函数(RBF)神经网络方法,取行星际扰动参数和GIC作为网络的输入—输出数据,利用空间扰动参数对GIC进行提前预测。结果显示:预测值和实际值的相关系数为0.96;将RBF神经网络方法应用到地磁感应电流的预测中,能取得较满意的效果,可为电网的减灾防灾提供依据。

To prevent the impact of the geomagnetically induced current （GIC） on the power grid, we can forecast the GIC by using the spatial parameters. The GIC can be forecast by using radial basis function （RBF） neural network with interplanetary disturbance parameter and GIC as the input and output of the network respectively. The result shows that correlation coefficient between the predicted value and the actual measured value is 0.96, indicating the satisfactory forecasting effect can be gained by application of the RBF neural network method in the forecast of GlC.