基于改进神经网络的电流互感器综合状态评估的研究
Research on Comprehensive State Assessment of Current Transformer Based on Improved Neural Network
杨鹏举 1王涛云 1杨恒 1孟垂攀 1张颢1
作者信息
- 1. 国网上海市电力公司金山供电公司,上海 200540
- 折叠
摘要
针对电流互感器长期在恶劣工况下运行造成准确度退化以及故障问题,设计一种电子式电流互感器在线状态综合评估系统,通过神经网络实现对互感器在线误差预测与故障诊断.针对现有基于神经网络的方法存在收敛速度慢、精度低等问题,提出一种基于改进的鲸鱼优化神经网络用于误差预测与故障诊断.通过非线性收敛因子提高鲸鱼优化的收敛速度;同时引入自适应惯性权重与模拟退火机制提高鲸鱼优化算法精度,避免陷入局部最优.通过基准函数测试、算例分析验证了方法有效性与可靠性.实验证明:所设计的电流互感器在线状态综合评估系统能有效地对电流互感器进行误差预测与故障诊断.
Abstract
For the accuracy degradation and fault caused by long-term operation of current transformers under harsh operating conditions,an electronic current transformer online condition comprehensive evaluation system is proposed to achieve online error prediction and fault diagnosis of transformers based on neural networks.With regard to the slow convergence and low accuracy of existing neural network-based methods,a neural network based on improved whale optimization is put forward for error prediction and fault diagnosis.The convergence speed of whale optimization is accelerated by nonlinear convergence factor,and meanwhile,adaptive inertia weights and simulated annealing mechanism are introduced to improve the accuracy of the whale optimization algorithm and avoid falling into local optimal.Benchmark function test and case analysis are conducted to verify the validity and reliability of the method.The experiment proves that designed the online state comprehensive evaluation system of current transformer designed can effectively perform error prediction and fault diagnosis of current transformer.
关键词
互感器误差预测/故障诊断/鲸鱼优化算法/人工神经网络Key words
transformer error prediction/fault diagnosis/whale optimization algorithm/BP neural network引用本文复制引用
基金项目
国家电网有限公司科技项目(SGSHJS00HBJS2103087)
出版年
2024
国家科技期刊平台
NSTL
万方数据