基于振动和线圈电流信号的低压断路器故障诊断研究  

Research on fault diagnosis of low-voltage circuit breaker based on vibration and coil current signals

作　　者：彭威 张钢[2] PENG Wei;ZHANG Gang(Guangzhou Metro Group Co.,Ltd.,Guangzhou 510335,China;School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]广州地铁集团有限公司,广东广州510335 [2]北京交通大学电气工程学院,北京100044

出　　处：《电工电能新技术》2025年第2期106-115,共10页Advanced Technology of Electrical Engineering and Energy

基　　金：北京市自然科学基金-丰台轨道交通前沿研究联合基金项目(L231027)。

摘　　要：为确保低压断路器在供电系统中工作的稳定性和可靠性,提出了一种基于遗传算法(GA)和深度置信网络(DBN)的故障诊断方法。该方法使用多物理域信息的特征作为网络的输入,对低压断路器进行故障诊断。首先,通过传感器采集低压断路器的振动信号和电流信号,并采用小波软阈值法进行降噪处理,有效避免了传统方法中特征信息的丢失。其次,提取两类信号时域和频域特征形成多物理域信息的特征矩阵,作为后续模型的输入。最后,将多物理域特征矩阵输入DBN神经网络进行网络微调,用GA遗传算法优化DBN网络的权值和阈值,提高故障诊断模型的准确性。实验结果表明,本研究所提模型的诊断精度达到98.46%,相较于仅使用振动信号或电流信号的诊断模型,分别提高了1.49%和6.79%,充分证明了基于多物理域的GA-DBN诊断模型在低压断路器故障诊断中的优异性能。To guarantee the stability and reliability of low-voltage circuit breakers within power supply systems,this paper proposes a fault diagnosis method that is based on a combination of Genetic Algorithm(GA)and Deep Belief Network(DBN).The proposed method employs features derived from multi-physical domain information as input to the network for the purpose of diagnosing faults in low-voltage circuit breakers.Firstly,vibration and current signals from the circuit breakers are collected via sensors and denoised using wavelet soft thresholding to preserve feature information effectively,avoiding losses typical in conventional methods.Subsequently,time-domain and frequency-domain features are extracted from both signal types to form a multi-physical domain feature matrix,which serves as input for subsequent modeling.The multi-physical domain feature matrix is then fed into a DBN neural network for fine-tuning,with GA optimizing the network’s weights and thresholds to enhance fault diagnosis accuracy.Experimental results demonstrate that the proposed model achieves a diagnostic accuracy of 98.46%.Compared to models using vibration or current signals alone,this represents an improvement of 1.49%and 6.79%,respectively.This fully demonstrates the superior performance of the multi-physical domain-based GA-DBN diagnostic model in fault diagnosis of low-voltage circuit breakers,further validating the criticality and practicality of this technology in the intelligent power supply system of subways.

关 键 词：振动信号 分合闸线圈电流 小波阈值 多物理域特征提取 深度置信网络 

分 类 号：TM561[电气工程—电器]