改进量子蝙蝠优化BiLSTM-KELM的电缆接头故障早期预警  

Cable joint fault early warning based on improved quantum bat algorithm optimizing BiLSTM-KELM

作　　者：严丹昭 陈晶 方遒 张瑞清 兰旺耀 廖一鹏[3] YAN Danzhao;CHEN Jing;FANG Qiu;ZHANG Ruiqing;LAN Wangyao;LIAO Yipeng(Fuzhou Power Supply Company,State Grid Fujian Electric Power Co.,Ltd.,Fuzhou 350009,China;Fuzhou Zhongxiang Electronic Information Technology Co.,Ltd.,Fuzhou 350026,China;College of Physics and Information Engineering,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]国网福建省电力有限公司福州供电公司,福建福州350009 [2]福建众想电子信息科技有限公司,福建福州350026 [3]福州大学物理与信息工程学院,福建福州350108

出　　处：《现代电子技术》2023年第14期93-99,共7页Modern Electronics Technique

基　　金：国家自然科学基金项目(62271149);国家自然科学基金项目(62271151);福建省自然科学基金项目(2019J01224);福州亿力电力工程有限公司配电工程分公司资助项目(RNFW2022GJT041013-Z)。

摘　　要：为提高对电缆中间接头故障的事先预知能力,文中提出一种基于改进量子蝙蝠算法优化BiLSTM-KELM模型的电缆中间接头故障早期预警方法。首先,采集电缆及接头的表层温度、环境温度、负荷电流的时间序列数据作为驱动,建立基于双向长短时记忆网络(BiLSTM)和核极限学习机(KELM)的接头中心温度预测模型;然后,构建非线性自适应旋转角量子旋转门以改进速度和位置的更新策略,并引入量子非转门实现较差个体的量子位置变异,用于预测模型参数的优化;最后,对正常工作接头进行温度预测和残差计算,使用概率分布拟合计算故障预警的残差阈值。实验结果表明,改进后的量子蝙蝠算法可以较好地逼近全局最优解,收敛效率高;优化后BiLSTM-KELM模型的预测精度得到有效提高,故障预警时间进一步提前,电缆接头故障的早期预警效果好。In order to improve the predictive ability of cable intermediate joint faults,a cable intermediate joint fault early warning method based on the improved quantum bat algorithm optimizing BiLSTM-KELM model is proposed.The time series data of surface temperature,ambient temperature and load current of the cables and their joints are collected as the driving force to establish a joint center temperature prediction model based on BiLSTM(bi-directional long short-term memory)-KELM(kernel-based extreme learning machine).A nonlinear adaptive rotating angle quantum turnstile is constructed to improve the speed and position updating strategy,and the quantum non-turnstile is introduced to realize the quantum position variation of poor individuals,which can be used to optimize the prediction model parameters.The temperature prediction and residual calculation of normal working joints are carried out,and the residual threshold of fault warning is calculated by means of probability distribution fitting.The experimental results show that the improved quantum bat algorithm can better approximate the global optimal solution and has high convergence efficiency,the prediction accuracy of the optimized BiLSTM-KELM model is effectively improved,and the fault warning time is further advanced.The early warning effect of cable joint faults is good.

关 键 词：量子蝙蝠算法 电缆接头 故障预警 双向长短时记忆网络 核极限学习机 温度预测 参数优化 残差计算 

分 类 号：TN919-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]