基于机液联合仿真模型的闸门启闭机故障注入及诊断研究  

Fault injection and diagnosis of gate hoist based on mechanical-hydraulic joint simulation model

作　　者：李浩宇 徐平[2] 铁瑛[1] 黄建章 LI Haoyu;XU Ping;TIE Ying;HUANG Jianzhang(College of Mechanical and Power Engineering,Zhengzhou University,Zhengzhou 450001,China;College of Water Resources and Transportation,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学机械与动力工程学院,郑州450001 [2]郑州大学水利与交通学院,郑州450001

出　　处：《振动与冲击》2024年第13期198-209,共12页Journal of Vibration and Shock

摘　　要：为了对含故障特征的弧形闸门系统运行信号进行评估和故障诊断研究,基于某弧形闸门试验平台检测数据,建立弧形闸门启闭机构的机液联合仿真模型,进行模型相似度分析和修正。通过基于闸门实时开度反馈信号闸门状态控制策略对闸门进行工作状态的控制,提高了仿真效率和稳定性,获得更符合实际的状态特征。在模型中注入液压系统与机械系统故障,分析故障对闸门运行信号产生的影响。为了充分体现故障特征信息,选取多个信号进行融合并调整权重分配,基于加权多通道数据融合方法解决了故障识别精度波动大的问题,为了提高网络模型的泛化能力,在网络模型中加入残差结构进行优化,基于迁移学习的卷积神经网络解决了故障识别精度低的问题。结果表明,构建的模型可以表现出弧形闸门工作过程中重要信号比如压力、流量、振动等信号特征的动态变化;使用改进后的加权多通道数据融合的基于残差结构优化的神经网络准确率达到了97.17%。Here,to evaluate and diagnose operating signals of a radial gate system with fault features,a mechanical-hydraulic joint simulation model of radial gate opening and closing mechanism was established based on detection data of a certain radial gate experimental platform to perform model similarity analysis and correction.By using a gate state control strategy based on real-time gate opening feedback signals,gate working state was controlled to improve simulation efficiency and stability,and obtain more realistic state features.Hydraulic system and mechanical system faults were injected into the model to analyze effects of faults on gate operation signals.In order to fully reflect fault feature information,multiple signals were selected for fusion and adjusting weight allocation.Based on the weighted multi-channel data fusion method,the large fluctuation problem of fault recognition accuracy was solved.In order to improve the generalization ability of the network model,residual structures were added for optimization.The convolutional neural network based on transfer learning was used to solve the problem of low fault recognition accuracy.The results showed that the constructed model can demonstrate feature dynamic changes of important signals of pressure,flow rate and vibration,etc.in operation process of radial gate;the correct rate of the neural network based on residual structure optimization using improved weighted multi-channel data fusion reaches 97.17%.

关 键 词：弧形闸门 机液联合仿真 故障诊断 迁移学习 

分 类 号：TV663.2[水利工程—水利水电工程] TV664.2