首页|基于电机电流数据的转台机械谐振检测方法

基于电机电流数据的转台机械谐振检测方法

Mechanical resonance detection of turntable based on motor current data

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
转台作为航空航天领域关键测试设备,在测试试验时可能出现谐振,造成测试过程的中断及不必要损失.针对传统转台谐振检测方法存在的依赖工程经验、易受控制信号干扰以及准确性不足等问题,提出了一种基于电机电流数据的转台机械谐振检测方法.首先,对三轴立式转台运行数据进行了分析,研究了转台运行数据的幅频特性,选用了受控制信号干扰小,且谐振特征更显著的电机电流信号作为检测对象;然后,构建了一个基于卷积神经网络(CNN)的转台谐振检测方法,设计了 CNN 结构与参数,进行了网络训练和学习;最终,利用不同工作模式下的转台运行数据,进行了谐振检测试验,并将检测试验结果与采用其他方法所得的结果进行了对比.研究结果表明:在转台谐振检测中,采用该方法所获得的准确率达到了 99.988％,优于采用其他对比方法获得的准确率,证明该方法适用于转台的机械谐振检测.
As a key testing equipment in the aerospace field,turntables may experience resonance during testing,causing interruptions and unnecessary losses in the testing process.Traditional turntable resonance detection method has problems such as dependence on engineering experience,susceptibility to control signal interference,and insufficient accuracy,resulting in false alarms and missed alarms.Aiming at the problems of traditional turntable resonance detection method,a mechanical resonance detection method for turntables based on motor current data was proposed.Firstly,the operating data of a three-axis vertical turntable was analyzed,and the amplitude frequency characteristics of the turntable operating data were studied.The motor current signal with less control signal interference and more significant resonance characteristics was selected as the detection object.Then,a turntable resonance detection method based on convolutional neural network(CNN)was constructed,and the CNN structure and parameters were designed,followed by network training and learning.The network structure and parameters were optimized through experiments to achieve better recognition results.Finally,resonance detection experiments were conducted using data from the turntable operating in different working modes,and the results were compared with support vector machine,long short-term memory neural network and gated recurrent neural network.The research results show that the accuracy of this method in detecting turntable resonance reaches 99.988％,which is better than other comparative methods,proving that this method is suitable for mechanical resonance detection of turntables.This method can be extended to the resonance detection of other equipment with motor.

mechanical vibrationvibration testoperating data of three-axis vertical turntablemotor currentconvolutional neural networks(CNN)frequency domain characteristics of convolution kernel

冯睿哲、马雅琼、闫斌斌

展开 >

西北工业大学 航天学院,陕西 西安 710072

航空工业北京长城航空测控技术研究所,北京 100176

机械振动 振动测试 三轴立式转台运行数据 电机电流 卷积神经网络 卷积核频域特征

航空科学基金资助项目

20200001053005

2024

10.3969/j.issn.1001-4551.2024.08.017

机电工程
浙江大学　浙江省机电集团有限公司

机电工程

CSTPCD北大核心
影响因子：0.785
ISSN：1001-4551
年,卷(期)：2024.41(8)
  • 18