摘要
电机和齿轮箱是船舶电力推进器重要的功能部件,其故障将严重影响推进器甚至整船的安全性。由于机电耦合作用的影响,当电机和齿轮箱同时发生故障时,故障信号信噪比低,故障特征存在交叉,为了诊断两者的耦合故障,提出1DCNN和2D-DCNN双分支卷积神经网络模型(TCNN),通过不同尺度的卷积核深入提取电流数据的全局特征和细节特征。在数据预处理方面,改进了二维灰度图构建方法以增强信号时间序列连续性,并在2D-DCNN通道中引入膨胀因子在不增加计算量的情况下挖掘信号中的全局信息,使用学习率指数衰减策略确保模型在迭代循环中稳定逼近最优解。试验结果表明,TCNN模型与其他模型相比具有更好的诊断性能,诊断准确率可达99.8%。同时在不同工作环境下,模型的诊断准确率都不低于98.5%,具有良好的适应性和鲁棒性。研究成果可为解决船舶电力推进器机电耦合故障的诊断问题提供新的思路和方法。
Motor and gearbox are essential functional components of marine electric thruster,and their faults seriously affect the safety of the thruster and even the whole ship.Due to the influence of electromechanical coupling,when simultaneous failures occur in both the motor and gearbox,the signal-to-noise ratio of the fault signal is low,and the fault characteristics are crossed.In order to diagnose the coupling fault between the two,a Two-Branch Convolutional Neural Network(TCNN)model of One-Dimensional Convolutional Neural Network(1DCNN)and Two-Dimensional Dilated Convolutional Neural Network(2D-DCNN)is proposed,which deeply extracts the global and detailed features of current data through convolution kernels of different scales.In terms of data preprocessing,the two-dimensional grayscale image construction method is improved to enhance the continuity of signal time series,and the expansion factor is introduced into the 2D-DCNN channel to extract the global information in the signal without increasing the amount of calculation.The exponential decay is used to ensure that the model can stably approximate the optimal solution in the iterative cycle.The experimental results show that the TCNN model has better diagnostic performance than other models,and its diagnostic accuracy can reach 99.8%.At the same time,the diagnostic accuracy rate of the model is not less than 98.5%in different working environments,which has good adaptability and robustness.The research results provide new ideas and methods for solving the problem of electromechanical coupling fault diagnosis of marine electricthrusters.
作者
姚诚武
盛晨兴
欧阳武
张雪琴
董小伟
YAO Chengwu;SHENG Chenxing;OUYANG Wu;ZHANG Xueqin;DONG Xiaowei(Wuhan University of Technology,School of Naval Architecture,Ocean and Energy Power Engineering,Wuhan 430063,China;Wuhan University of Technology,State Key Laboratory of Maritime Technology and Safety,Wuhan 430063,China;Wuhan University of Technology,Reliability Engineering Institute,National Engineering Research Center for Water Transport Safety,Wuhan 430063,China;Wuhan University of Technology,School of Transportation and Logistics Engineering,Wuhan 430063,China;China Merchants Cruise Shipbuilding Co.,Ltd.,Nantong 226100,Jiangsu,China)
出处
《船舶工程》
CSCD
北大核心
2024年第7期58-65,73,共9页
Ship Engineering
基金
湖北省自然科学基金杰出青年项目(2023AFA098)。
关键词
船舶电力推进器
故障诊断
双分支卷积神经网络
电机
齿轮箱
机电耦合故障
marine electric thruster
fault diagnosis
two-branch convolutional neural network
motor
gearbox
electromechanical coupling fault
