摘要
随着配电自动化技术不断发展,对配电开关运行可靠性提出了更高的要求,然而当前配电开关晨操时健康状态评估准确率低,无法保障电网安全、稳定、可靠运行。基于此,提出一种基于瞪羚算法优化长短期记忆网络的配电开关晨操健康状态评估方法。首先,分析影响开关健康状态因素;其次,使用最大信息系数把跟配电开关健康状态最相关的因素提取出来,作为长短期记忆网络输入;最后,对长短期记忆网络进行改进,设计瞪羚算法长短期记忆网络的配电开关健康状态评估模型,通过瞪羚算法优化长短期记忆网络超参数、注意力机制重点关注重要特征的方式提高模型的训练精度和效率。实验验证,所提方法有效提升了配电开关晨操时健康状态评估能力,保证设备安全可靠运行。
With the continuous development of distribution automation technology,higher requirements have been put forward for the reliability of distribution switch operation.However,the current accuracy of health status assessment during morning operation of distribution switches is low,which cannot guarantee the safe,stable,and reliable operation of the power grid.In view of this problem,a method for assessing the health status of distribution switches during morning operation by using the gazelle algorithm to optimize long short term memory network is proposed.Firstly,the factors that affect the health status of the switch is analyzed;secondly,the maximum information coefficient is used to extract the factors most related to the health status of the distribution switch as the inputs for the long short term memory network;finally,a health status assessment model for distribution switches by using the gazelle algorithm to support long short term memory network is designed,so as to optimize the long short term memory network.The accuracy and efficiency of model’s training are improved by optimizing the hyperparameters of the long short term memory network and focusing on important features under attention mechanism through the gazelle algorithm.Experimental verification shows that the method proposed effectively improves the ability to assess the health status of distribution switches during morning operation,ensuring the safety and reliability of equipment operation.
作者
李浩然
王子滔
LI Haoran;WANG Zitao(Shenzhen Power Supply Bureau Co.,Ltd.,Shenzhen 518000,China)
出处
《机械与电子》
2025年第6期25-30,共6页
Machinery & Electronics
基金
深圳供电局有限公司科技项目(09000020240301030900108)。
关键词
配电开关
晨操
健康状态评估
瞪羚算法
长短期记忆网络
distribution switches
morning operation
health status assessment
gazelle algorithm
long short term memory network
