基于改进的小波神经网络模型在覆冰预测中的应用研究

Application Research of Modified Wavelet Neural Network Model in Ice Covered Prediction

输电线路覆冰厚度数据对输电线路冰灾防治具有重要意义。线路覆冰达到一定厚度后,电线张力和杆塔荷载会达到危险水平,需要采取相应的措施。建立覆冰厚度预测模型,可以预测某一时间点的覆冰厚度值,为运行单位提供决策参考。利用线路的覆冰历史数据,选择小波神经网络建立覆冰厚度预测模型,并利用共轭梯度算法代替传统的训练算法,显著提高了建模速度。预测结果表明,这种模型具有较好的容错能力,并满足预测精度。

It＇s very important to obtain the transmission line＇s ice covered thickness for ice disaster prevention. When the ice covered thickness reaches a certain value, wire tension and tower load will reach a dangerous level, so the corresponding precautions must be taken. A prediction model of the ice thickness can predict the ice covered thickness at a certain time and it will provide the operation department with a reference for the decision making. This paper builds the model by use of partial historical data and wavelet neural network, and uses conjugate gradient method for training the network. The modeling speed is significantly increased. The prediction results show that the model is of higher fault tolerance capability and prediction accuracy.