基于机器学习的输电走廊强对流临近预警技术

Early warning technology for severe convective approaches in transmission corridor based on machine learning

输电走廊上的强对流天气会给电力生产、供应带来严重的威胁,从而影响居民的正常用电需求。针对此问题,研究了基于机器学习的输电走廊强对流临近预警技术。该技术选取有效反映输电走廊强对流预警的数据类型,构建数据特征并对样本数据进行预处理,再利用卷积与池化操作实现数据的特征识别。同时采用CL多小波融合方法将细节融入目标数据库以增加物理信息量,为后续预警处理提供更为准确的数据,并使用了交叉相关算法来解决图像的追踪问题。通过实际的算例分析,验证了所提算法的有效性,其预测精度可达93.64%。

The severe convective weather in the transmission corridor will pose a serious threat to power production and supply and affect the normal power demand of residents.To solve this problem,this paper studies the early warning technology of severe convection in transmission corridor based on machine learning.Select the data types that effectively reflect the strong convection early warning in the transmission corridor,construct the data characteristics,and then preprocess the sample data.The feature recognition of data is realized through convolution and pooling operation.CL multi wavelet fusion method is used to integrate the details into the target database,increase the amount of physical information,provide more accurate data for subsequent early warning processing,and use cross⁃correlation algorithm to solve the problem of image tracking.Through the analysis of practical examples,the effectiveness of the proposed algorithm is verified,its prediction accuracy can reach 93.64%.