首页>
外国专利>
Fault Diagnosis System For Rotating Device Using Deep Learning and Wavelet Transform
Fault Diagnosis System For Rotating Device Using Deep Learning and Wavelet Transform
展开▼
机译:深度学习和小波变换的旋转设备故障诊断系统
展开▼
页面导航
摘要
著录项
相似文献
摘要
The present invention relates to a fault diagnosis system for a rotating body using deep running and wavelet transform, and more particularly, to a fault diagnosis system for a rotating body, which comprises a sensing unit for acquiring a point-in-time signal for an input load applied to the rotating body or an output load outputted through the rotating body; A converter for converting the time-point-based signal into time and frequency components and converting the time-frequency signals into an image form; A learning unit for processing the converted image through the conversion unit by processing the learning input data of the deep learning module; And a diagnostic unit for diagnosing whether the rotor is broken according to a feature extracted from the signal based on a result of the learning unit; And a control unit. According to another aspect of the present invention, there is provided a fault diagnosis method for a rotating body using deep running and wavelet transform, the method comprising: a first step of obtaining a point-of-view signal for an input load applied to the rotating body or an output load outputted through the rotating body; A second step of converting the time-point-based signal into time and frequency components and transforming the time-frequency signals into an image form; A third step of processing the image converted through the second step by processing the learning input data of the deep learning module; And a fourth step of diagnosing whether or not the rotating body is faulty according to characteristics extracted from the signal based on the result of the third step. And a control unit. Thus, the limited actual data of the rotating body is converted into a two-dimensional image through wavelet transformation and is learned through the deep learning module, thereby extracting the signal characteristic of the rotor failure itself, thereby improving the reliability of the diagnosis. In addition, since the data converted into the two-dimensional image is accumulated in a predetermined reference time based on time and frequency, the preprocessing process for data quantification can be omitted.
展开▼