基于定子电流和电磁转矩双信号融合的齿轮故障智能诊断  

Intelligent Gear Fault Diagnosis for Dual-signal Fusion Based on the Stator Current and the Electromagnetic Torque

作　　者：李巍[1] 袁响东 陈伟 刘军 LI Wei;YUAN Xiangdong;CHEN Wei;LIU Jun(College of Electronics and Information,Tongji University,Shanghai 201804;Shanghai STEP Electric Corporation,Shanghai 201801;Hangzhou Yuhang District Power Supply Company,State Grid Zhejiang Electric Power Co.,Ltd.,Hangzhou 311100)

机构地区：[1]同济大学电子与信息工程学院,上海201804 [2]上海新时达电气股份有限公司,上海201801 [3]国网浙江省电力有限公司杭州市余杭区供电公司,杭州311100

出　　处：《电气工程学报》2024年第3期248-256,共9页Journal of Electrical Engineering

摘　　要：在电机驱动的齿轮传动系统中,电机本体具有传感器的特性,因此可以通过电机的定子电流、电磁转矩信号来进行齿轮故障分析,由于受转速和负载转矩的影响,使得故障诊断结果的准确率较低。针对此问题,提出一种基于双信号融合与反向传播神经网络相结合的齿轮故障诊断方法。对电机齿轮传动系统一体化建模,进行电机齿轮传动系统联合仿真。对齿轮的不同故障进行模拟,得到电机侧定子电流和电磁转矩的故障信号,采用双树复小波变换来分析齿轮故障频段信号,提取故障特征量,建立了丰富的齿轮故障样本库。搭建反向传播神经网络并提出改进的自适应学习率算法,实现了对齿轮断齿、磨损故障的精确分类。为了验证所提方法的有效性,搭建齿轮故障试验平台,对相应齿轮故障进行诊断。结果表明,所提方法能够在不同转速和负载转矩条件下准确辨识齿轮的故障类型,相较于只采用定子电流和电磁转矩中一种信号对齿轮进行故障诊断,该方法准确率更高。In the motor driven gear transmission system,the motor body has the characteristics of sensors,so the stator current and electromagnetic torque signal of the motor can be used to analyze the gear fault.Due to the influence of the speed and load torque,the accuracy of the fault diagnosis results is low.Aiming at this problem,a gear fault diagnosis method based on dual signal fusion and back propagation neural network is proposed.The integrated modeling of motor gear transmission system is carried out,and the co-simulation of motor gear transmission system is carried out.By simulating different gear faults,the fault signals of stator current and electromagnetic torque of motor side are obtained.The gear fault frequency signals are analyzed by double-tree complex wavelet transform,fault characteristics are extracted,and a rich gear fault sample database is established.The back propagation neural network is built and an improved adaptive learning rate algorithm is proposed to accurately classify gear broken teeth and wear faults.In order to verify the effectiveness of the proposed method,a gear fault experimental platform is built to diagnose the corresponding gear faults.The results show that the proposed method can accurately identify gear fault types under different speed and load torque conditions.Compared with only using a signal of stator current and electromagnetic torque for gear fault diagnosis,the proposed method has a higher accuracy.

关 键 词：齿轮故障 传动系统 神经网络 双树复小波变换 智能诊断 

分 类 号：TM56[电气工程—电器]