Deep learning-based pressure sensor fault diagnosis method and device

Abstract

The invention relates to a deep learning-based pressure sensor fault diagnosis method and system. The method comprises the steps of obtaining a fault signal of a to-be-detected pressure sensor; performing short-time Fourier transform and empirical mode decomposition on the fault signal respectively; according to the harmonic number of the frequency spectrum and the information entropies of a plurality of modal functions obtained in the empirical mode decomposition process, determining the modal number of variational mode decomposition; performing variational mode decomposition on the fault signal according to the mode number to obtain an envelope spectrum of the fault signal; and inputting the frequency spectrum, the marginal spectrum and the envelope spectrum of the fault signal into a trained fault identification model to obtain the fault classification and the fault reason of the pressure sensor to be detected. According to the method, the time domain, the time-frequency domain, the energy spectrum and other characteristics of the fault are obtained through modal decomposition and Fourier transform, and then the fault type is identified by using the identification model based on deep learning, so that the universality and the flexibility of identification are improved.

Description

technical field [0001] The invention belongs to the field of pressure sensor fault diagnosis identification and machine learning, and in particular relates to a fault diagnosis method and device for a pressure sensor based on deep learning. Background technique [0002] With the maturity and popularization of 5G and other communication technologies and the continuous development of various smart devices, high requirements are put forward for the measurement accuracy, real-time performance, reliability and self-confirmation of sensors in various smart devices. Furthermore, as the basis of various measurement and control systems or intelligent systems, the structure of pressure sensors is becoming more and more complex, and the corresponding model conditioning circuits will increase accordingly, and the possibility of failure will increase; if a failure occurs, it may affect the operation of the entire system. The consequences are disastrous; in recent years, the fault diagnos...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problems of complex models, poor versatility and poor flexibility of the existing pressure sensor fault diagnosis technology, a method for fault diagnosis of pressure sensors based on deep learning is provided in the first aspect of the present invention, comprising the following steps: obtaining the The fault signal of the pressure sensor; the short-time Fourier transform and empirical mode decomposition are respectively carried out on the fault signal to obtain the frequency spectrum and the marginal spectrum of the fault signal; according to the number of harmonics of the frequency spectrum, and the The information entropy of multiple modal functions obtained in the process of empirical mode decomposition is used to determine the modal number of variational modal decomposition; the fault signal is subjected to variational modal decomposition according to the modal number to obtain the described The envelope spectrum of the fault signal; input the frequency spectrum, marginal spectrum and envelope spectrum of the fault signal into the trained fault identification model to obtain the fault classification and fault cause of the pressure sensor to be tested

Images