Improved diagnosis method for complex faults of rotating machinery with deep sparse autoencoder network

Abstract

To improve the diagnosis method for complex faults of rotating machinery with deep sparse autoencoder network, 1) obtain the vibration signals of c kinds of working conditions in the normal state and different fault states of the rotating machinery respectively, and obtain d groups of time-domain vibration signal samples; 2) for each Fourier transform of samples to obtain preprocessed signal samples; 3) Construct a diagnostic sample set and use it as a training sample set; 4) Establish a composite fault diagnosis model for rotating machinery and obtain the connection weights of the deep sparse autoencoder network and bias parameters; 5) Obtain the softmax classifier model; 6) Perform fast Fourier transform and select test samples; 7) The test samples are used as the input of the trained improved deep sparse autoencoder network, and the test samples are deeply learned , to perform feature extraction to obtain the characteristic signal of the test sample; 8) use the test feature information as the matching feature of the test sample to obtain the composite fault diagnosis result of the rotating machinery to be tested; improve the diagnostic accuracy and efficiency.

Description

technical field [0001] The invention belongs to the technical fields of mechanical fault diagnosis and computer artificial intelligence, and in particular relates to an improved diagnosis method for compound faults of rotating machinery in deep sparse autoencoder networks. Background technique [0002] In recent years, the equipment in the industrial system has become increasingly large-scale, continuous, complex, high-speed and automated, which has also become the main feature of modern large-scale enterprise production. In today's era, the development of industrial technology also puts forward higher requirements for the safety and reliability of industrial production processes, especially in the pillar industries of the national economy. If the failure of production equipment cannot be prevented in time, once a production accident occurs, it will cause great economic losses, and even cause casualties and environmental pollution. [0003] The common features of these indu...

AI Technical Summary

Problems solved by technology

These processes need to make full use of human experience and knowledge in signal processing and fault diagnosis. The recognition accuracy of fault diagnosis depends on the quality of feature extraction, which greatly reduces the adaptive ability of diagnostic methods.

In addition, the existing diagnostic methods for single components, subsystems, and subunits are difficult to find out the correlation between monitoring data, which inevitably increases the risk of misdiagnosis and missed diagnosis in composite fault diagnosis

Images