摘要
以提升电子元件故障诊断能力,设计基于能量谱特征分析的耦合电路故障识别系统。高速扫描采集电路采集耦合电路运行信号,利用数字通道将耦合电路运行信号输入到故障识别模块内。使用能量谱检测计算耦合电路能量谱密度,将该密度数值转换成时频图谱,卷积神经网络建立电路故障识别模型,进行故障识别,并生成故障图谱和故障告警结果,将结果通过人机交互模块内呈现给用户。实验结果表明:该系统识别耦合电路故障时输出结果均方误差低,具备较高的样本有效性;计算耦合电路能量谱密度较为精准,可有效识别其串联电弧故障并发出故障告警,具备较好的应用性。
In order to improve the fault diagnosis ability of electronic components,a coupling circuit fault identification system based on energy spectrum feature analysis is designed.The high-speed scanning acquisition circuit collects the operation signal of the coupling circuit,and uses the digital channel to input the operation signal of the coupling circuit into the fault identification module.The energy spectrum density of coupling circuit is calculated by energy spectrum detection,and the value of the density is converted into time-frequency spectrum.The convolution neural network is used to establish the circuit fault identification model for fault identification,generate the fault spectrum and fault alarm results,and present the results to the user through the human-computer interaction module.The experimental results show that the system has low mean square error and high sample validity when identifying coupling circuit faults;The calculation of energy spectral density of coupling circuit is more accurate,which can effectively identify the series arc fault and send out fault alarm,and has good application.
作者
邹丹
高扬
ZOU Dan;GAO Yang(Xi’an mingde institute of technology,Shaanxi.Xi’an 710024,China;Chang an university,Shaanxi.Xi’an 710064,China)
出处
《自动化与仪器仪表》
2022年第4期124-128,共5页
Automation & Instrumentation
基金
陕西省自然科学联合基金项目“基于惯性导航与视觉同步定位构图的综合定位与控制算法”研究成果(2019JLP-07)。
关键词
能量谱
卷积神经网络
耦合电路
故障系统
时频图谱
energy spectrum
convolutional neural network
coupling circuit
fault system
time frequency spectrum
