融合模糊评价与极限学习机的配电线路台风灾损预测

为了提升配电网线路应对台风的防灾减灾能力,提出一种台风影响下配电线路的灾损预测方法.充分考虑台风登陆位置对目标地区影响程度的复杂性,利用模糊评价方法评估台风登陆距离的影响模糊程度.并在此基础上结合台风风级、风速和风圈半径等属性作为预测模型的输入特征,构建基于极限学习机(ELM)的灾损预测模型,对台风影响下的目标地区配电线路跳闸、断线、倒杆以及断杆进行预测.通过对近几年台风历史数据的实验验证表明,本算法针对四类灾损能获得较高的预测精度;相比现有算法,对登陆距离的影响程度进行模糊评价极大提升了预测模型的精度,实现20倍以上的性能提升.算法预测结果可为台风过境时配电线路的灾害应急处理和灾后重建工作提供重要参考依据.

Damage Prediction for Distribution Line under Typhoon Disaster Based on Extreme Learning Machine and Fuzzy Evaluation

To improve the damage prediction accuracy for distribution line under typhoon disaster, ELM based on fuzzy algorithm method for electric power damage caused by a typhoon has been proposed. The fuzzy processing method is used to transform the typhoon landing point information into membership function due to it is complex and difficult to confirm. Then, the damage prediction model of ELM is established based on the disaster such as tripping and distribution poles falling. Finally, the model accuracy is measured by RMSE (the root-mean-square error) of the sample test. The experiment result shows that the proposed ELM model outperforms traditional ELM algorithm and other neural network algorithms in terms of forecasting accuracy. It provides an effective basis for disaster emergency treatment and reconstruction of distribution lines under typhoon disaster.