基于小波变换的自动翻译系统电路故障自动诊断系统研究  

Research on automatic fault diagnosis system of automatic translation system based on wavelet transform

作　　者：杨宁[1] Yang Ning(Xi’an Fanyi University,Xi’an 710105,China)

机构地区：[1]西安翻译学院,西安710105

出　　处：《自动化与仪器仪表》2024年第10期29-32,共4页Automation & Instrumentation

摘　　要：为了使自动翻译系统可以持续稳定地提供英语翻译服务,提出一种基于小波变换的电路故障自动诊断方法,提高英语自动翻译系统的可靠性。该方法一方面使用经验小波变换的方法提高故障特征的提取效果;另一方面通过多分支ResNet构建智能分类模型,基于经验小波变换所提取的故障特征,实现对电路故障的自动诊断。结果表明,基于经验小波变换方法提取的故障特征,在同一故障类型上聚类性良好,在不同的故障类型之间具有较高的区分度。利用经验小波变换提取的故障特征,多分支ResNet模型在对Sallenkey带通滤波电路13种故障类型的自动诊断中,诊断准确度接近100%,相较于WT+CNN和EEMD+Entropy+ELM等前沿故障诊断方法,提高了约0.57%和4.72%,自动诊断性能优越,更适合用于英语翻译系统的日常维护。In order to make the automatic translation system provide English translation service continuously and stably,an automatic circuit fault diagnosis method based on wavelet transform is proposed to improve the reliability of the automatic translation system.On the one hand,the method uses empirical wavelet transform to improve the fault feature extraction effect.On the other hand,the intelligent classification model is constructed by multi-branch ResNet,and the automatic fault diagnosis is realized based on the fault features extracted by the empirical wavelet transform.The results show that the fault features extracted based on the empirical wavelet transform method have good clustering on the same fault type and high differentiation among different fault types.Using the fault features extracted by empirical wavelet transform,the multi-branch ResNet model achieves a diagnostic accuracy close to 100%in the automatic diagnosis of 13 fault types in Sallenkey bandpass filter circuit.Compared with cutting-edge fault diagnosis methods such as WT+CNN and EEMD+Entropy+ELM,Improved by about 0.57%and 4.72%,the automatic diagnosis performance is superior,and it is more suitable for daily maintenance of English translation system.

关 键 词：小波变换 电路故障 自动诊断 残差网络 英语翻译 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]