摘要
为了实现更加准确同时鲁棒性更强的电力采集终端故障诊断,设计基于数据驱动的配电网电力采集终端故障诊断模型。基于瞬时有功电流-瞬时无功电流法实施配电网电力采集终端的谐波测量,实现谐波数据采集。在谐波数据采集中,由于存在栅栏效应,采集获得的信号会由于丢失一些特征量而与实际谐波信号出现误差。应用加窗插值FFT算法来弥补栅栏效应,实现谐波修正,使故障诊断更加准确。利用ISSA实施深度信念网络的参数寻优,设计ISSA-DBN故障诊断模型,将修正后的谐波数据作为模型输入数据,实现基于数据驱动的配电网电力采集终端故障诊断。实验测试结果为设计模型在四种不同的故障种类下均取得了较高的辨识匹配值,其中通信故障的辨识匹配值最高,达到了0.991,随着噪音水平的降低,设计模型的故障辨识匹配值虽然也有所下降,但下降幅度相对较小,表明该模型具有较好的鲁棒性。
In order to achieve more accurate and robust fault diagnosis of power collection terminals,a data-driven fault diagnosis model for power collection terminals in distribution networks is designed.Based on the instantaneous active current instantaneous reactive current method,harmonic measurement of distribution network power acquisition terminals is implemented to achieve harmonic data collection.In harmonic data collection,due to the fence effect,the collected signal may have errors with the actual harmonic signal due to the loss of some characteristic quantities.Applying the windowed interpolation FFT algorithm to compensate for the fence effect and achieve harmonic correction,making fault diagnosis more accurate.Using ISSA to implement parameter optimization of deep belief networks,design an ISSA-DBN fault diagnosis model,and use the corrected harmonic data as input data to achieve data-driven fault diagnosis of power acquisition terminals in distribution networks.The experimental test results showed that the design model achieved high identification matching values under four different types of faults,with the highest identification matching value of 0.991 for communication faults.As the noise level decreased,although the fault identification matching value of the design model also decreased,the decrease was relatively small,indicating that the model has good robustness.
作者
陆沈敏
徐爱蓉
张辉
寿佳珏
陈珊
LU Shenmin;XU Ai’rong;ZHANG Hui;SHOU Jiajue;CHEN Shan(QingPu Power Supply Company,State Grid Shanghai Municipal Electric Power Company,Shanghai 201799,China)
出处
《自动化与仪器仪表》
2025年第7期57-60,共4页
Automation & Instrumentation
基金
国网上海市电力公司科技项目(520934230003)。
