摘要
基于舰船发动机燃油电磁阀驱动电路,分析电磁阀电流特性与故障情况下的电流特征,发现燃油电磁阀的电流波动,对电磁阀故障具有重要影响。因此,利用小波包分解技术重构电流信号,提取电流信号的频带幅值,将其作为舰船发动机燃油电磁阀不同故障的特征向量,将该特征向量输入多输入层卷积神经网络中,经过训练、测试的多输入层卷积神经网络可以准确输出电磁阀的不同故障类型。实验结果表明,该方法可准确提取舰船发动机燃油电磁阀故障信号中的各类状态特征,诊断出电磁阀正常、弹簧断裂和阀芯卡死的故障类型,可靠性高。
A real-time fault diagnosis method for Marine engine fuel solenoid valve is proposed to realize the real-time diagnosis of different types of solenoid valve faults.Based on the driving circuit of fuel oil solenoid valve of ship engine,the current characteristics of the solenoid valve and the current characteristics under the condition of failure are analyzed.It is found that the current fluctuation of fuel oil solenoid valve has an important effect on the failure of solenoid valve.Therefore,the current signal,using wavelet packet decomposition technique is to extract frequency amplitude of current signal,as a ship engine fuel solenoid valve different fault feature vector,the feature vector convolution neural network input multiple input layer,after training,testing of multiple input layer convolution neural network can accurately output different fault types of electromagnetic valve.The experimental results show that this method can accurately extract all kinds of fault features from the fault signals of the fuel solenoid valve of ship engine,and diagnose the fault types of normal solenoid valve,broken spring and stuck valve core,with high reliability.
作者
孙月秋
SUN Yue-qiu(Bohai Ship-building Vocational College,Huludao 125001,China)
出处
《舰船科学技术》
北大核心
2022年第3期121-124,共4页
Ship Science and Technology
基金
2019年船舶技术专业委员会科研课题。
关键词
发动机燃油
电磁阀故障
小波包分解
卷积神经网络
engine fuel
solenoid valve failure
wavelet packet decomposition
convolutional neural network
